BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Department of Electrical & Computer Engineering - ECPv6.15.18//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Department of Electrical & Computer Engineering
X-ORIGINAL-URL:https://ece.northeastern.edu
X-WR-CALDESC:Events for Department of Electrical & Computer Engineering
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20220313T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20221106T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20230312T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20231105T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20241103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230721T140000
DTEND;TZID=America/New_York:20230721T153000
DTSTAMP:20260416T232904
CREATED:20230718T175222Z
LAST-MODIFIED:20230718T175222Z
UID:6433-1689948000-1689953400@ece.northeastern.edu
SUMMARY:Daniel Uvaydov's PhD Dissertation Defense
DESCRIPTION:Title: Real-Time Spectrum Sensing for Inference and Control \nCommittee Members: \nProf. Tommaso Melodia (Advisor) \nProf. Kaushik Choudhury \nProf. Francesco Restuccia \nAbstract: \nThrough growing cellular innovations\, the usage and congestion of the wireless spectrum is increasing at incredible speeds. High demand and limited supply pose a resource issue known as the “spectrum crunch”. With the high diversity of users sharing a large portion of the spectrum to request and receive diverse services\, spectrum coordination becomes very difficult. Large scale device synchronization for spectrum coordination requires high overhead and more wireless transmissions further reducing spectrum resources. However\, by monitoring the spectrum\, otherwise known as spectrum sensing\, we can develop mechanisms where users can opportunistically take action based on the current state of the spectrum\, without need for direct coordination between devices. Spectrum sensing can enable the next generation of wireless applications ranging from opportunistic spectrum access to cognitive radio networks. The key unaddressed challenges of spectrum sensing are that (i) it requires very extensive and diverse datasets; (ii) it has to be performed with extremely low latency over varying bandwidths and must guarantee strict real-time processing constraints; (iii) its underlying algorithms need to be extremely accurate\, and flexible enough to work with different wireless bands and protocols to find application in real-world settings. This dissertation focuses on addressing these challenges in multiple wireless applications by utilizing Deep Learning (DL) techniques as the main vehicle of spectrum sensing for both inference and control. Algorithmic spectrum sensing has generally been model-based which limits its performance in diverse settings and environments\, for this reason we explore data-driven spectrum sensing algorithms. Mainly\, this work takes a holistic approach to address spectrum sensing problems from multiple directions with the overarching goal of developing the core building blocks for the next generation of intelligent\, AI-driven\, efficient spectrum sharing systems. By leveraging mechanisms such as data augmentation\, channel attention\, voting\, and segmentation we are able to push beyond the capabilities of existing DL techniques and create generalizable spectrum sensing algorithms. Furthermore we deploy different spectrum sensing solutions in real testbeds for over the air evaluations and applicable proof-of-concepts. The contributions of this work includes (i) multiple datasets and implementations for DL enabled spectrum sensing with applications in radio frequency and underwater; (ii) a method for tackling the core issue of dataset generation in supervised learning algorithms for spectrum sensing via a novel data augmentation technique; (iii) a study into one of the first ever semi-unsupervised approaches for wideband multi-class spectrum sensing.
URL:https://ece.northeastern.edu/event/daniel-uvaydovs-phd-dissertation-defense/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230720T130000
DTEND;TZID=America/New_York:20230720T140000
DTSTAMP:20260416T232904
CREATED:20230711T180015Z
LAST-MODIFIED:20230711T180015Z
UID:6425-1689858000-1689861600@ece.northeastern.edu
SUMMARY:Qing Jin's PhD Dissertation Defense
DESCRIPTION:Title:Decoupling Efficiency-Performance Optimization for Modern Neural Networks \nDate: \n7/20/2023 \nCommittee Members: \nYanzhi Wang (Advisor); Prof. David Kaeli; Prof. Sunil Mittal; Prof. Jennifer Dy \nAbstract: \nDeep learning has achieved remarkable success in a variety of modern applications\, but this success is often accompanied by inefficiency in terms of storage and inference speed\, which can hinder their practical use on resource-constrained hardware. Developing highly efficient neural networks that maintain high prediction accuracy is crucial and challenging. This dissertation explores the potential for simultaneously achieving high efficiency and high prediction accuracy in neural networks\, and can be broadly divided into three sections. (1) In Section One\, we explore the implementation of highly efficient generative adversarial networks (GANs) capable of generating high-quality images within a predefined computational budget. The key challenge lies in identifying the optimal architecture for the generative model while simultaneously preserving the quality of the generated images from the compressed model\, despite its reduced computational cost. To achieve this\, we propose a novel neural architecture search (NAS) algorithm and a new knowledge distillation technique. (2) In Section Two\, we explore the challenge of quantizing discriminative models without relying on high-precision multiplications. To address this issue\, we present an innovative approach to determine the optimal fixed-point formats for both weights and activations based on their statistical properties. Our results demonstrate that high accuracy in quantized neural networks can be achieved without the need for high-precision multiplications. (3) In Section Three\, we delve into the challenge of training neural networks for innovative computing platforms\, specifically processing-in-memory (PIM) systems. Through a detailed mathematical derivation of the backward propagation algorithm\, we facilitate the training of quantized models on these platforms. Additionally\, through a thorough theoretical analysis of training dynamics\, we ensure convergence and propose a systematic solution for quantizing neural networks on PIM systems.
URL:https://ece.northeastern.edu/event/qing-jins-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230629T170000
DTEND;TZID=America/New_York:20230629T173000
DTSTAMP:20260416T232904
CREATED:20230626T213009Z
LAST-MODIFIED:20230626T213036Z
UID:6407-1688058000-1688059800@ece.northeastern.edu
SUMMARY:Zifeng Wang's PhD Dissertation Defense
DESCRIPTION:Title: Effective and Efficient Continual Learning \nCommittee Members:\nProf. Jennifer Dy (Advisor)\nProf. Stratis Ioannidis\nProf. Yanzhi Wang \nAbstract:\nContinual Learning (CL) aims to develop models that mimic the human ability to learn continually without forgetting knowledge acquired earlier. While traditional machine learning methods focus on learning with a certain dataset (task)\, CL methods adapt a single model to learn a sequence of tasks continually. \nIn this thesis\, we target developing effective and efficient CL methods under different challenging and resource-limited settings. Specifically\, we (1) leverage the idea of sparsity to achieve cost-effective CL\, (2) propose a novel prompting-based paradigm for parameter-efficient CL\, and (3) utilize task-invariant and task-specific knowledge to enhance existing CL methods in a general way. \nWe first introduce our sparsity-based CL methods. The first method\, Learn-Prune-Share (LPS)\, splits the network into task-specific partitions\, leading to no forgetting\, while maintaining memory efficiency. Moreover\, LPS integrates a novel selective knowledge sharing scheme\, enabling adaptive knowledge sharing in an end-to-end fashion. Taking a step further\, we present Sparse Continual Learning (SparCL)\, a novel framework that leverages sparsity to enable cost-effective continual learning on edge devices. SparCL achieves both training acceleration and accuracy preservation through the synergy of three aspects: weight sparsity\, data efficiency\, and gradient sparsity. \nSecondly\, we present a new paradigm\, prompting-based CL\, that aims to train a more succinct memory system that is both data and memory efficient. We first propose a method that learns to dynamically prompt (L2P) a pre-trained model to learn tasks sequentially under different task transitions\, where prompts are small learnable parameters maintained in a memory space. We then improve L2P by proposing DualPrompt\, which decouples prompts into complementary “General” and “Expert” prompts to learn task-invariant and task-specific instructions\, respectively. \nFinally\, we propose DualHSIC\, a simple and effective CL method that generalizes the idea of leveraging task-invariant and task-specific knowledge. DualHSIC consists of two complementary components that stem from the so-called Hilbert Schmidt independence criterion (HSIC): HSIC-Bottleneck for Rehearsal (HBR) lessens the inter-task interference and HSIC Alignment (HA) promotes task-invariant knowledge sharing. \nComprehensive experimental results demonstrate the effectiveness and efficiency of our methods over the state-of-the-art methods on multiple CL benchmarks.
URL:https://ece.northeastern.edu/event/zifeng-wangs-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230626T083000
DTEND;TZID=America/New_York:20230626T093000
DTSTAMP:20260416T232904
CREATED:20230624T220848Z
LAST-MODIFIED:20230624T220848Z
UID:6402-1687768200-1687771800@ece.northeastern.edu
SUMMARY:Deniz Unal's PhD Proposal Review
DESCRIPTION:Title:\nSoftware-Defined Underwater Acoustic Networks \nCommittee Members:\nProf. Tommaso Melodia (Advisor)\nProf. Stefano Basagni\nProf. Kaushik Chowdhury\nDr. Emrecan Demirors \nAbstract:\nThe exploration\, monitoring\, and understanding of oceans play a crucial role in addressing climate change\, overseeing underwater pipelines\, and preventing maritime warfare attacks. To achieve these significant objectives\, it is vital to utilize networks of cost-effective and flexible underwater devices capable of efficiently collecting and transmitting information to the shore. However\, the progress of underwater networks heavily relies on underwater acoustic modems\, which currently face limitations such as low data rates and inflexible hardware designs\, limiting their usability to specific scenarios. To overcome these limitations\, we propose a modular software-defined acoustic networking platform built on the Zynq system-on-chip architecture that can be easily deployed in a compact form factor. Our platform distinguishes itself from existing solutions in several ways. Firstly\, it possesses the capability to adapt to varying conditions by adjusting protocol parameters at all layers of the networking stack. Secondly\, it achieves high data rate connections\, particularly over short distances. Additionally\, it seamlessly integrates with other sub-sea platforms\, including underwater drones. We demonstrate the capabilities and the performance of our platform with tasks\, such as channel estimation and characterization\, establishing high data rate Orthogonal Frequency-Division Multiplexing (OFDM) links\, and running third-party software to implement JANUS standard. In addition\, we introduce the enabling technologies for the development and implementation of underwater networks. These technologies facilitate the establishment of connectivity between underwater networks and the shore\, as well as the integration of modems with underwater vehicles. Lastly\, we provide a demonstration of the algorithmic development conducted on our platform. We mainly consider high-rate\, wideband\, adaptive links and perform experimental evaluations at sea. In particular\, we demonstrate multicarrier communications with mobile platforms with the presence of Doppler and compare the performance of forward error correction methods\, and demonstrate dataset recording for artificial intelligence research.
URL:https://ece.northeastern.edu/event/deniz-unals-phd-proposal-review/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230623T100000
DTEND;TZID=America/New_York:20230623T110000
DTSTAMP:20260416T232904
CREATED:20230606T193237Z
LAST-MODIFIED:20230606T193237Z
UID:6395-1687514400-1687518000@ece.northeastern.edu
SUMMARY:Cooper Loughlin's PhD Dissertation Defense
DESCRIPTION:“Deep Generative Models for High Dimensional Spatial and Temporal Data Analysis” \nCommittee Members:\nProf. Vinay Ingle (Advisor)\nDr. Dimitris Manolakis\nProf. Purnima Ratilal-Makris \nAbstract:\nData analysis and exploitation in practical applications is challenging when observations are the result of many interacting natural and man-made phenomena. We address two important problems for which traditional methods of analysis are insufficient. One problem of practical interest is the identification of particular materials from remotely sensed hyperspectral imagery. This is traditionally accomplished by comparing image pixel spectra to those from a known material library. Such techniques are limited by spectral variability\, background interference\, and imperfect compensation of atmospheric components. Established methods address these limitations with statistical techniques. Simple probability models result in tractable methods; however\, analyses are limited by errors due\, in particular\, due to false alarms. \nAnalysis of complex time series is another challenging problem\, particularly when data are high dimensional. This arises in air quality monitoring\, where atmospheric concentration measurements of multiple pollutants are taken over time. Two analysis goals in this context are forecasting and anomaly detection. Both tasks are enabled by an accurate model for the temporal dynamics and interaction between pollutants. Air quality data are complex due to long term temporal dependencies\, non-linear dependence between pollutants\, and missing observations. Traditional multivariate time series analysis approaches\, such as the vector autoregression and linear dynamical system models\, fail to capture those characteristics necessary for a sufficient probabilistic model. \nWe use deep generative models to develop practical solutions that address these problems. This is made possible through the application of deep latent variable models. The modeling approach follows the philosophy that complex data can typically be explained by simpler underlying factors of variation. Variational autoencoders (VAEs) are deep latent variable models that emulate data generation by transforming simple\, low dimensional\, latent random vectors through a deep neural network. VAEs are trained to produce samples that resemble the training data\, thus capturing a manifold on which complex data are distributed. This philosophy is extended to time series data\, where we consider sequences of latent vectors. \nWe utilize VAEs develop a flexible generative model for hyperspectral imagery. Based on that model\, we develop a novel material identification framework which localizes target material spectra along the manifold. Through experiments on real data\, we show that the \ac{VAE} approach is better able to reject false alarms from materials with similar spectra when compared to established methods alone. We additionally develop a novel dynamical \ac{VAE} model for time series of air quality data. Using that model\, we develop practical methods for computing forecast distributions using Monte Carlo integration. We evaluate forecast distributions against real air quality data and demonstrate the ability to predict temporal dynamics and forecast uncertainty. The primary contribution of this work is to develop practical solutions to challenging data analysis problems through the use of deep generative models.
URL:https://ece.northeastern.edu/event/cooper-loughlins-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230620T130000
DTEND;TZID=America/New_York:20230620T140000
DTSTAMP:20260416T232904
CREATED:20230522T212041Z
LAST-MODIFIED:20230522T212041Z
UID:6354-1687266000-1687269600@ece.northeastern.edu
SUMMARY:Chang Liu's PhD Dissertation Defense
DESCRIPTION:“Unleashing the Potential of Transfer Learning for Visual Applications” \nCommittee Members:\nProf. Raymond Fu (Advisor)\nProf. Sarah Ostadabbas\nProf. Zhiqiang Tao \nAbstract:\nThe recent flourish of deep learning in various tasks is largely accredited to the rich and accessible labeled data. Nonetheless\, massive supervision remains a luxury for many real-world applications. Further\, the domain shift problem has also seriously impeded large-scale deployments of deep-learning models. As a remedy\, Transfer learning aims at improving the performance of target learners on target domains by transferring the knowledge contained in different but related source domains. In this way\, the dependence on a large number of target domain data can be reduced for constructing target learners. \nIn this dissertation research\, I investigate two major problems in transfer learning\, domain adaptation (DG) and domain adaptation (DA)\, on various visual applications. (1) The challenge of DG lies in an over-simplified assumption\, that is\, the source and target data are independent and identically distributed (i.i.d.) while ignoring out-of-distribution (OOD) scenarios commonly encountered in practice. This issue is common in visual applications such as object recognition\, hyperparameter optimization\, and face recognition. We propose algorithms that are specifically designed for each task\, such as metric learning\, adversarial regularization\, feature disentanglement\, and meta-learning. (2) DA can be considered a special case of DG with unlabeled target data available. The major challenge is how to align the labeled source and unlabeled target data. We delve into the applications of image recognition and video recognition and propose algorithms to ensure domain-wise discriminativeness and class-wise closeness across domains. Experiments show that the proposed algorithms outperform the state-of-the-art methods on the commonly-used benchmark datasets.
URL:https://ece.northeastern.edu/event/chang-lius-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230620T080000
DTEND;TZID=America/New_York:20230620T170000
DTSTAMP:20260416T232904
CREATED:20230624T221028Z
LAST-MODIFIED:20230624T221028Z
UID:6404-1687248000-1687280400@ece.northeastern.edu
SUMMARY:Alfred P. Navato's PhD Dissertation Defense
DESCRIPTION:Title:\nEnabling Anomaly Detection in Complex Chemical Mixtures Through Multimodal Data Fusion \nDate:\n6/26/2023 \nTime:\n10:00:00 AM \nLocation:\nSH 210\, \nCommittee Members:\nProf. Mueller (Advisor)\nProf. Erdogmus\nProf. Ioannidis\nProf. Onnis-Hayden \nAbstract:\nRecently innovations in machine learning and data processing are increasingly tied to ensuring useability and interpretability when these methods are applied within end-user domains.  One societally important example of such a domain is management and operations of water infrastructure in cities\, where data collection is currently costly and limited\, enabling analytics have the potential to generate real impact for urban communities\, and correctness of results is critical to protect human and environmental health.  This dissertation holistically considers issues of generalizability\, transferability\, and applicability of a range of data fusion and machine learning approaches across end-user domains within the context of solution building for improved real-time management of wastewater infrastructure.  The first chapter provides an overview of the challenges associated with anomaly detection within the wastewater field and reviews the performance of various anomaly detection techniques implemented in other disciplines.  The second chapter discusses the barriers and opportunities in cross-disciplinary pollination of data fusion techniques.  The third chapter presents development of an unsupervised approach facilitating quantitative characterization of the complex background which is wastewater\, necessary to be able to implement any automated operational interventions.  The fourth chapter develops an approach for cost-minimization/information-maximization design of a sensor to facilitate specifically detection of chemical anomalies (defined as inflow events that might compromise wastewater treatment facilities) by using machine learning and feature selection techniques to minimize the number of input signals needed to achieve reasonable accuracies.  Together the third and fourth chapters provide a clear\, explainable\, actionable pathway forward in envisioning next generation wastewater infrastructure\, demonstrating novel and impactful use of data fusion and machine learning techniques in a real-world context.
URL:https://ece.northeastern.edu/event/alfred-p-navatos-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230613T173000
DTEND;TZID=America/New_York:20230613T191500
DTSTAMP:20260416T232904
CREATED:20230522T181116Z
LAST-MODIFIED:20230617T021216Z
UID:6356-1686677400-1686683700@ece.northeastern.edu
SUMMARY:Ignorance Is Bliss: A Career Retrospective
DESCRIPTION:  \nDean Gregory D. Abowd will present his SIGCHI Lifetime Research Award Acceptance Lecture \nDate: Tues.\, June 13\, 2023 \nTime: 5:30 to 7:15 PM\, reception following Dean Abowd’s talk \nPlace: In-Person and Livestream\nBostonCHI meeting at Northeastern University in ISEC Auditorium (102 ISEC)\, and reception in ISEC Atrium \nRegistration is appreciated but not required. View BostonCHI for more information. \nPresentation Abstract: In 1988\, as a graduate student grappling to find a research identity\, Gregory D. Abowd accidentally discovered the field of Human Computer Interaction (HCI). Over the past 35 years\, he pursued a passion for applying the tools and techniques of computing to uncover how the human experience with technology can be understood and transformed. That leap into HCI was just the first of a number of leaps of faith. Abowd’s career has been a series of shifting research agendas\, each one inspired by some life events. In all cases\, he was buoyed by a bevy of talented and supportive colleagues\, advisors and advisees alike\, who gave him the courage to jump into a research topic that he didn’t know much about. That “ignorance” has allowed him to be more fearless than he had the right to be. In this talk\, Abowd will reflect on his professional journey\, hoping to inspire others to dispel fear of the unknown and unlock their potential. Life\, like research\, is best when shared with others whom you can respect and befriend. \n—————————————— \nGregory D. Abowd\, dean of the College of Engineering and professor of electrical and computer engineering at Northeastern University\, has received the Lifetime Research Award from the Association for Computing Machinery’s (ACM) Special Interest Group on Computer-Human Interaction (SIGCHI). The award is presented to individuals for “the best\, most fundamental\, and influential research contributions to the study of human-computer interaction (HCI)” and is awarded for a lifetime of innovation and leadership. \n 
URL:https://ece.northeastern.edu/event/ignorance-is-bliss-a-career-retrospective/
LOCATION:102 ISEC\, 360 Huntington Ave\, 102 ISEC\, Boston\, MA\, 02115\, United States
GEO:42.3377335;-71.0869121
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=102 ISEC 360 Huntington Ave 102 ISEC Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 102 ISEC:geo:-71.0869121,42.3377335
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230605T110000
DTEND;TZID=America/New_York:20230605T123000
DTSTAMP:20260416T232904
CREATED:20230522T211405Z
LAST-MODIFIED:20230522T211405Z
UID:6345-1685962800-1685968200@ece.northeastern.edu
SUMMARY:Can Qin's PhD Dissertation Defense
DESCRIPTION:“Unveiling the Power of Transfer Learning in Data-Driven AI” \nCommittee Members:\nProf. Raymond Fu (Advisor)\nProf. Octavia Camps\nProf. Huaizu Jiang \nAbstract:\nThe big data stands as a cornerstone of deep learning\, which has significantly improved a wide range of machine learning and computer vision tasks. Despite such a great success\, data collection is time-consuming and costly\, considering manual efforts and privacy restrictions. Transfer learning is a promising direction toward data-efficient AI by leveraging acquired data and pre-trained models as guidance. This dissertation focus on the feature and model transfer across different domains and tasks\, which can be roughly summarized into three sections. \n(1) Section One focuses on Unsupervised Domain Adaptation (UDA) without any labels in the target domain. The technical challenge of UDA is the distribution mismatch across domains. I have presented a hierarchical alignment model as the solution. \n(2) Section Two extends UDA into semi-supervised domain adaptation (SSDA) with minimal target-domain labels\, which is useful and effortless to acquire. To avoid overfitting toward labeled data\, I have proposed structural regularization verified on different classification benchmarks. \n(3) Section Three mainly explores the model transfer\, including teacher-student knowledge distillation and heterogeneous models infusion with a high potential for model compression and enhancement.
URL:https://ece.northeastern.edu/event/can-qins-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230602T110000
DTEND;TZID=America/New_York:20230602T120000
DTSTAMP:20260416T232904
CREATED:20230508T193647Z
LAST-MODIFIED:20230508T193647Z
UID:6319-1685703600-1685707200@ece.northeastern.edu
SUMMARY:Cheng Gongye's PhD Proposal Review
DESCRIPTION:“Hardware Security Vulnerabilities in Deep Neural Networks and Mitigations” \nCommittee Members:\nProf. Yunsi Fei (Advisor)\nProf. Xue Lin\nProf. Xiaolin Xu \nAbstract:\nOver the past decade\, Deep Neural Networks (DNNs) have revolutionized numerous fields. With the increasing deployment of DNN models in security-sensitive and mission-critical applications\, such as autonomous driving\, ensuring the security and privacy of DNN inference is of paramount importance. \nThis Ph.D. dissertation investigates two primary hardware security attack vectors: fault attacks and side-channel attacks. Fault attacks compromise the integrity of a targeted application by intentionally disrupting the computation or injecting faults on parameters. Side-channel attacks exploit information leakage from the application execution through physical parameters such as power consumption\, electromagnetic emanations\, and timing to retrieve secrets\, thereby breaching confidentiality. \nFor fault attacks\, we demonstrate a power-glitching fault injection attack on FPGA-based DNN accelerators in cloud environments. The attack exploits vulnerabilities in the shared power distribution network and leverages time-to-digital converter (TDC) sensors for precise fault injection timing\, and results in model misclassification\, an integrity compromise on the targeted application. We propose a lightweight defense framework for detecting and mitigating adversarial bit-flip attacks induced by RowHammer on DNNs. This framework employs a dynamic channel-shuffling obfuscation scheme and a logits-based model integrity monitor\, offering negligible performance loss. This framework effectively protects various DNN models from RowHammer attacks without any retraining or model structure modifications. \nFor side-channel attacks\, we present a floating-point timing side channels attack to reverse-engineer multi-layer perceptron (MLP) model parameters in software implementations. This attack successfully recovers DNN parameters\, weights and biases. \nRegarding ongoing research\, we observe that previous studies often focus on academic prototypes\, resulting in limited applicability. To bridge these gaps\, we select the AMD-Xilinx DPU\, one of the most advanced DNN accelerators to date\, to conduct the analysis. We propose a side-channel attack that utilizes electromagnetic emissions to extract parameters. Furthermore\, we propose a comprehensive fault analysis of quantized DNN models by simulations and discuss potential mitigation strategies.
URL:https://ece.northeastern.edu/event/cheng-gongyes-phd-proposal-review/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230526T123000
DTEND;TZID=America/New_York:20230526T133000
DTSTAMP:20260416T232904
CREATED:20230522T211528Z
LAST-MODIFIED:20230522T211528Z
UID:6347-1685104200-1685107800@ece.northeastern.edu
SUMMARY:Guillem Reus Muns' PhD Dissertation Defense
DESCRIPTION:“AI for communication and sensing in RF environments” \nCommittee Members:\nProf. Kaushik Chowdhury (Advisor)\nProf. Stratis Ioannidis\nProf. Hanumant Singh \nAbstract:\nThe recent growth of Internet of Things (IoT)\, as well as other new\nrevolutionary applications utilizing wireless spectrum are leading the way towards the realization of next-generation wireless systems that jointly utilize communications and sensing. However\, such systems offer many degrees of freedom\, and optimizing them for a specific task is difficult to accomplish with deterministic and classical approaches. For this reason\, data-driven and AI-based methods have been pursued actively by the research community\, as they are able to find solutions that often come close to or exceed the performance of the deterministic counterparts with fractional design complexity. This thesis presents\, through real systems and with experimental validation\, our progressive efforts in four broad areas\, where AI enables the operation of aerial and terrestrial systems that combine sensing and communications. The following key use cases with distinct contributions are investigated: \ni) Sensing-aided communications for air and ground systems. First\, we present a UAV communication method that defines constellation points in space that map to transmitter frequency bands and are detected at the Base Station using millimeter wave sensors. Second\, we explore alternative vehicle-to-infrastructure mmWave beamforming methods\, leveraging a) vehicle position and velocity estimation using in-band standard compliant 802.11ad radar and b) camera images and GPS location information. \nii) Signal classification using communication signals\, where we propose a) a UAV classification method using uniquely UAV-transmitted signals and b) an RF fingerprinting technique that improves class separation by combining triplet loss with regular classification techniques. \niii) ‘SenseORAN’\, a revolutionary architectural design that aims to reuse the cellular infrastructure for sensing purposes in order to address spectrum access challenges in the CBRS band. This is enabled by Open Radio Access Network (O-RAN)\, a cellular architecture concept that promotes virtualized RANs where disaggregated components are connected via open interfaces and supports intelligent controllers running custom logic. iv) ‘AirFC’\, an over-the-air computation method that implements fully connected neural networks inference leveraging multi-antenna wireless systems.
URL:https://ece.northeastern.edu/event/guillem-reus-muns-phd-dissertation-defense/
LOCATION:Admissions Visitor Center (West Village F)
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230526T090000
DTEND;TZID=America/New_York:20230526T100000
DTSTAMP:20260416T232904
CREATED:20230522T211659Z
LAST-MODIFIED:20230522T211659Z
UID:6350-1685091600-1685095200@ece.northeastern.edu
SUMMARY:Yuezhou Liu's PhD Proposal Review
DESCRIPTION:Committee Members:\nProf. Edmund Yeh (Advisor)\nProf. Stratis Ioannidis\nProf. Lili Su\nProf. Carlee Joe-Wong \nAbstract:\nSignificant advances in edge and mobile computing capabilities enable machine learning to occur at geographically diverse locations in networks\, e.g.\, cloud\, edge\, and mobile devices. The training data needed in those learning tasks may not be fully generated locally. Moreover\, some promising distributed learning paradigms enable devices to collaboratively train a model\, which requires communication among the devices for exchanging necessary information. Thus\, optimizing network strategies for the transmission/exchange of ML/AI ingredients (e.g.\, input data\, model parameters\, gradients) is important for facilitating efficient in-network distributed ML. While there exist many works that use ML to optimize network operation strategies\, few works study optimized networks that boost ML performance. This dissertation tries to fill the gap by studying several network optimization problems for distributed ML. Different from classic network optimization problems for data delivery or edge computing that optimize energy consumption\, delay\, throughput\, etc.\, we also pay attention to ML-related metrics such as model accuracy and training convergence time. \nWe first propose an experimental design network paradigm\, wherein learner nodes train possibly different ML models via consuming data streams generated by data source nodes over a network. We formulate this problem as a social welfare optimization problem in which the global objective is defined as the sum of experimental design objectives of individual learners\, and the decision variables are the data transmission strategies subject to network constraints. We show that\, assuming Bayesian linear regression models and Poisson data streams in steady state\, the global objective is continuous DR-submodular\, which enables the design of efficient approximate algorithms with approximation guarantees. We will further extend our framework to incorporate more practical ML applications\, such as ML with arbitrary nonlinear models. \nThe second half of this dissertation studies network optimization for Federated learning (FL)\, a distributed paradigm for collaboratively learning models without having clients disclose their private data. We propose to use caching for improving FL efficiency concerning the total model training time for convergence. Instead of having all clients download the latest global model from a parameter server\, we select a subset of clients to access\, with smaller delays\, a somewhat stale global model stored in caches. We propose CacheFL — a cache-enabled variant of FedAvg\, and provide theoretical convergence guarantees in the general setting where the local data is imbalanced and heterogeneous. With this result\, we determine the caching strategies that minimize total wall-clock training time at a given convergence threshold for both stochastic and deterministic communication/computation delays.
URL:https://ece.northeastern.edu/event/yuezhou-lius-phd-proposal-review/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230522T103000
DTEND;TZID=America/New_York:20230522T113000
DTSTAMP:20260416T232904
CREATED:20230522T211916Z
LAST-MODIFIED:20230522T211916Z
UID:6352-1684751400-1684755000@ece.northeastern.edu
SUMMARY:Mengting Yan's PhD Dissertation Defense
DESCRIPTION:“Integrated Circuit Design Methods for Temperature-based Hardware Trojan Detection and Parametric Frequency Division in Next-Generation Systems-on-a-Chip” \nCommittee Members:\nProf. Marvin Onabajo (Advisor)\nProf. Yong-bin Kim\nProf. Yunsi Fei \nAbstract:\nNew needs for next-generation systems-on-a-chip (SoC) are emerging as the trend of globalization in the semiconductor industry becomes increasingly ubiquitous and the demand for low-power Internet-of-Things (IoT) devices continues to soar. Among various research directions\, this dissertation focuses on enhancing hardware security and on providing low-noise frequency sources for next-generation SoCs. Within this scope\, the described research addresses the challenge to improve on-chip anomaly detection capabilities\, and separately lays a foundation for the design of circuits to reduce the phase noise of on-chip oscillators. \nIn the first part of this dissertation\, an on-chip temperature-based Hardware Trojan (HT) detection system is introduced. The approach to detect inserted HTs relies on thermal profiling of the circuit under test (CUT) and side-channel analysis of the obtained temperature data. On-chip electrothermal coupling is modeled as part of a simulation technique that associates local thermal activities with circuit-level power consumption using a standard electrical simulator. To monitor the thermal profiles on chips with high sensitivity to local temperature changes as well as to enhance the resilience to flicker noise\, a fully-differential temperature sensor equipped with a chopping mechanism has been designed in 130-nm complementary metal-oxide-semiconductor (CMOS) technology\, which has a sensitivity of 840 V/◦C. The simulated temperature sensor output in the presence of noise and process variations is quantized by an analog-to-digital converter (ADC) model and processed using principal component analysis (PCA)\, which allows to determine the minimum detectable Trojan power and the design requirements for the on-chip ADC. With a modeled 8-bit ADC\, simulations of the HT detection system reveal a detection rate of 100% with a Trojan power down to 2.4 μW within the thermal profile of a CUT consuming 508 μW. A prototype 8-bit 1 MS/s successive approximation register (SAR) ADC for such a system was designed in 130-nm CMOS technology\, fabricated\, and tested. The measured effective number of bits (ENOB) is 7.27 bits up to the Nyquist frequency\, with a power consumption of 103.2 μW from a 1.2 V supply. Furthermore\, a 3-step analog calibration loop has been designed to compensate for the voltage offsets within the sensor circuits in the presence of device mismatches and process-temperature variations. The calibration loop settles within 300 μs to complete the offset calibration\, such that the input-referred offset has a standard deviation of 5.86 μV based on Monte Carlo simulations. \nIn the second part of this dissertation\, the on-chip realization of a parametric frequency divider (PFD) is explained. The low-power 2:1 frequency division at sub-6 GHz plays a critical role in on-chip phase noise reduction systems that exhibit nonlinear operations\, indicating promise for future integration into radio frequency (RF) SoCs. In particular\, the first current-driven PFD with an output frequency of 2.4 GHz is introduced\, which consists of three major blocks: (1) a custom PFD driver stage with a buffer to ease input driving\, (2) a purely passive PFD core with inductor-capacitor (LC) resonators\, and (3) an output driving stage with embedded band-pass filtering that suppresses undesirable output harmonics. A prototype PFD chip was fabricated in standard 65-nm CMOS technology\, and the corresponding measurement results are presented to characterize the performance of the new PFD. The minimum required supply voltage for the PFD driver is 1.4 V with an input frequency of 4.8 GHz\, whereas the PFD has an operating frequency range from 4.5 GHz to 5.1 GHz with a supply voltage of 1.5 V. To the best of the author’s knowledge\, the proposed PFD is the first on-chip CMOS implementation for sub-6 GHz applications\, which balances the trade-offs among frequency range\, power consumption\, and chip area constraints.
URL:https://ece.northeastern.edu/event/mengting-yans-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230428T130000
DTEND;TZID=America/New_York:20230428T150000
DTSTAMP:20260416T232904
CREATED:20230426T174141Z
LAST-MODIFIED:20230426T174141Z
UID:6293-1682686800-1682694000@ece.northeastern.edu
SUMMARY:Balaji Sundareshan's MS Thesis Defense
DESCRIPTION:“Cross-View Action Recognition using Transformers” \nCommittee Members:\n1. Prof. Octavia Camps (Advisor)\n2. Prof. Mario Sznaier\n3. Prof. Huaizu Jiang \nAbstract:\nCross-view action recognition (CVAR) seeks to recognize a human action when observed from a previously unseen viewpoint. This is a challenging problem since the appearance of action changes significantly with the viewpoint. Applications of CVAR include surveillance and monitoring of assisted living facilities where is not practical or feasible to collect large amounts of training data when adding a new camera. In this thesis\, we propose a method to perform cross-view action recognition from 2D skeleton data using Transformers. First\, we understand the interpretability of the basline network and its submodules by visualizing the saliency map. Next\, we integrate Transformers at different parts of the network for both single-clip and multi-clip and understand the impact on the performance. In the end\, we also discuss the necessity of pretraining sub-modules in the network and their impact on the performance.
URL:https://ece.northeastern.edu/event/balaji-sundareshans-ms-thesis-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230426T160000
DTEND;TZID=America/New_York:20230426T170000
DTSTAMP:20260416T232904
CREATED:20230426T174309Z
LAST-MODIFIED:20230426T174309Z
UID:6295-1682524800-1682528400@ece.northeastern.edu
SUMMARY:Rui Huang's MS Thesis Defense
DESCRIPTION:“Sputter Deposition and Characterization of Highly Textured BixTe1-x Thin Films” \nCommittee Members:\nProf. Nian-Xiang Sun (Advisor)\nProf. Marvin Onabajo\nProf. Yongmin Liu \nAbstract:\nThe discovery of topological insulators (TIs) provides a direction for scientists to understand the Quantum Spin Hall Effect (QSHE) and Spin-Orbit Coupling (SOP)  in condensed matter physics. After a decade\, people found that after the introduction of magnetism into TI\, the Time Reversal Symmetry (TSR) is broken\, producing Magnetic Topological Insulators (MTI). Meanwhile\, with the deposition of TI on the Magnetic Insulator (MI)\, the Spin-Orbit Torque was found in TI/MI structures. Introducing dopants into TI is another method to produce MTI. Mn-doped\,  Cr-doped\, and Ni-doped TI thin films have been explored recently. Thus\, the 3D TI\, Bi2Te3\, and MTI\, Ni: Bi2Te3\, thin film-based materials have been applied to some energy-efficient spintronic devices. However\, according to the Bi-Te phase diagram\, Bi2Te3 is one of the Bi-Te family. The narrow range of the Bi2Te3 phase is a challenge for people to deposit the correct phase on the InP (111) wafer due to the potential effect of defects.  In this Master thesis\, the textured BixTe1-x and Ni-doped BixTe1-x thin films are deposited on the InP (111) substrate through the RF Magnon Sputtering Tool with a Te capping layer under different deposition powers\, temperatures\, and post-annealing time. After the X-ray diffraction measurement on three samples with various conditions\, the textured Bi8Te7\, Bi8Te9\, and Ni: Bi8Te7 thin films are concluded based on the comparison between the theoretical XRD results with the experimental ones.
URL:https://ece.northeastern.edu/event/rui-huangs-ms-thesis-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230425T130000
DTEND;TZID=America/New_York:20230425T143000
DTSTAMP:20260416T232904
CREATED:20230420T223701Z
LAST-MODIFIED:20230420T223701Z
UID:6273-1682427600-1682433000@ece.northeastern.edu
SUMMARY:Rashida Nayeem's PhD Dissertation Defense
DESCRIPTION:“Human control of objects with nonlinear internal dynamics: Predictability as primary objective” \nLocation:\nEgan Research Ctr 206 \nCommittee Members:\nProf. Dagmar Sternad (Advisor)\nProf. Eduardo Sontag\nProf. Mario Sznaier\nDr. David Lin (Massachusetts General Hospital & Harvard Medical School) \nAbstract:\nHumans physically interact with complex objects in numerous daily activities. An example is picking up a cup of coffee where interaction forces arise between the hand and the sloshing liquid. For successful actions\, error corrections based on real-time sensed information are insufficient\, hence humans need to predict and preempt the evolving dynamics. Our previous work on the transport of a “cup of coffee” showed that humans seek to make the interaction dynamics simple\, i.e.\, predictable. Extending from previous work\, this thesis used a virtual paradigm where the “cup of coffee” was simplified to a cup with a ball sliding inside\, retaining the challenges of “a cup filled with coffee”: underactuation and nonlinearity. A series of experiments examined human strategies in different contexts to demonstrate that predictability is a control priority. The first experimental and modeling study examined how subjects explored and prepared the 2D cup-and-ball system prior to continuous interaction. Results showed that subjects converged to a small set of initial conditions that shortened initial transients\, enabling subjects to reach a more predictable steady state faster. Two follow-up studies examined the role of visual and haptic information and revealed that despite suboptimal exploration of the solution space\, subjects increased predictability of hand object interactions. System identification showed that visual information enabled subjects to simplify input-output behavior via appropriate object preparation. When deprived of haptic information subjects still achieved increased predictability but sacrificed orbital stability. A final study extended this basic paradigm to a clinical application to investigate if these insights could help in assessment of motor impairment after stroke in this functionally relevant ‘self-feeding’ task. To facilitate testing in a clinic\, a real-life 3D device was custom-developed where individuals after stroke moved a cup with a rolling ball inside on a table. Our theory-based predictability metric proved highly sensitive to quantify the degree of motor impairment after stroke. Taken together\, this thesis elucidated principles of human motor control in a complex interactive task. The insights have significant applications in clinical testing and may also inform robot manipulation of this understudied movement challenge.
URL:https://ece.northeastern.edu/event/rashida-nayeems-phd-dissertation-defense/
LOCATION:206 Egan\, 360 Huntington Ave\, 206 Egan\, Boston\, MA\, 02115\, United States
GEO:42.3376753;-71.0888734
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=206 Egan 360 Huntington Ave 206 Egan Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 206 Egan:geo:-71.0888734,42.3376753
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230424T110000
DTEND;TZID=America/New_York:20230424T123000
DTSTAMP:20260416T232904
CREATED:20230420T223436Z
LAST-MODIFIED:20230420T223740Z
UID:6277-1682334000-1682339400@ece.northeastern.edu
SUMMARY:Cunzheng Dong's PhD Proposal Review
DESCRIPTION:“Acoustically Actuated Magnetoelectric Antennas for VLF Communication and Magnetic Sensing” \nCommittee Members:\nProf. Nian Sun (Advisor)\nProf. Yongmin Liu\nProf. Hossein Mosallaei \nAbstract:\nSince the discovery of strong magnetoelectric (ME) coupling in two-phase ME laminate composites\, strain-mediated ME heterostructures have attracted a great deal of attention from academic and industrial research groups for their potential usage in magnetic sensors\, voltage tunable inductors\, magnetic tunable filters\, and miniaturized mechanical antennas\, etc. Acoustically actuated ME antennas have recently been demonstrated as a promising solution for very low frequency (VLF) communications and magnetic fields detection\, for their 2-3 orders of reduced dimensions\, outstanding sensitivity at resonance\, and robust immunity to electrical interferences than conventional electric antennas. Their performance and noise analysis are deeply investigated and discussed in this proposal review. \nFirstly\, A portable VLF communication system using one pair of ME antennas operating at their electromechanical resonance (EMR) is presented. The measured near-field radiation pattern reveals ME antennas are equivalent to magnetic dipole antennas. The magnetic field radiated by the ME transmitter has been predicted along with distance from near-field to far-field. The measured magnetic field distribution coincided well with the prediction\, and the maximum communication distance of 120 m has been achieved by single antenna unit. Antenna arrays are widely used as an effective approach to enhance radiation field intensity. By tunning all the driving signal for each antenna unit at the same frequency and in phase\, the total radiation field strength has been linearly enhanced by one order with 12 antenna arrays. Furthermore\, nonlinear antenna modulation (NAM) has also been successfully demonstrated on the ME antennas. \nSecondly\, a Metglas/Quartz based ME resonator as magnetic sensor for reception of VLF magnetic signals is presented. Metglas is a highly permeable magnetostrictive material which can effectively concentrate the magnetic fields. Moreover\, the high magnetostriction and low coercivity of Metglas can generate a distinct strain change in response to subtle magnetic fields. Piezoelectric single crystal Quartz is often used as electronic oscillators due to their extremely high Q factor with low noise and high stability. The combined properties of these two materials provide ME sensors an extremely high sensitivity and low magnetic noise of less than 10 fT at the EMR frequency. The VLF signal reception capability of the proposed ME sensor was also compared with a conventional VLF loop antenna and the PZT-5A based ME sensor. \nLastly\, a compact and sensitive system was developed to characterize the magnetomechanical properties\, such as the saturation magnetostriction\, piezomagnetic coefficient\, delta-E effect and magnetomechanical coupling factor of magnetic thin films. These magnetomechanical properties are critical in determining the performance of ME antennas. For saturation magnetostriction and piezomagnetic coefficient measurement\, a high precision optical probe was harnessed to measure the deflection of the magnetic thin film/Si cantilever due to strain change induced by domain rotation. The same cantilever samples were used for delta-E effect and magnetomechanical coupling factor characterization\, the DC bias magnetic field induced cantilever resonance frequency shift was used for calculating the change of elastic modulus.
URL:https://ece.northeastern.edu/event/cunzheng-dongs-phd-proposal-review/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230420T080000
DTEND;TZID=America/New_York:20230420T170000
DTSTAMP:20260416T232904
CREATED:20230420T223901Z
LAST-MODIFIED:20230420T223901Z
UID:6279-1681977600-1682010000@ece.northeastern.edu
SUMMARY:Chenghao Wang's MS Thesis Defense
DESCRIPTION:“Legged Walking on Inclined Surfaces” \nCommittee Members:\nProf. Alireza Ramezani(Advisor)\nProf. Miriam Leeser\nProf. Bahram Shafai \nAbstract:\nThe main contributions of this MS Thesis are centered around taking steps towards successful multi-modal demonstrations using Northeastern’s legged-aerial robot\, Husky Carbon. This work discusses the challenges involved in achieving multi-modal locomotion such as trotting-hovering and thruster-assisted incline walking and reports progress made towards overcoming these challenges. Animals like birds use a combination of legged and aerial mobility\, as seen in Chukars’s wing-assisted incline running (WAIR)\, to achieve multi-modal locomotion. Chukars use forces generated by their flapping wings to manipulate ground contact forces and traverse steep slopes and overhangs. Husky’s design takes inspiration from birds such as Chukars. This MS thesis presentation outlines the mechanical and electrical details of Husky’s legged and aerial units. The thesis presents simulated incline walking using a high-fidelity model of the Husky Carbon over steep slopes of up to 45 degrees.
URL:https://ece.northeastern.edu/event/chenghao-wangs-ms-thesis-defense/
LOCATION:532 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230413T160000
DTEND;TZID=America/New_York:20230413T170000
DTSTAMP:20260416T232904
CREATED:20230405T174232Z
LAST-MODIFIED:20230405T174232Z
UID:6246-1681401600-1681405200@ece.northeastern.edu
SUMMARY:Hussein Hussein’s PhD Dissertation Defense
DESCRIPTION:“Parametric Circuits for Enhanced Sensing and RF Signal Processing” \nCommittee Members: \nProf. Cristian Cassella (Advisor) \nProf. Marvin Onabajo \nProf. Matteo Rinaldi \nProf. Andrea Alù \nAbstract: \nMassive deployments of wireless sensor nodes (WSNs) that continuously detect physical\, biological or chemical parameters are needed to truly benefit from the unprecedented possibilities opened by the Internet‑of‑Things (IoT). Just recently\, new sensors with higher sensitivities have been demonstrated by leveraging advanced on‑chip designs and microfabrication processes. Yet\, WSNs using such sensors require energy to transmit the sensed information. Consequently\, they either contain batteries that need to be periodically replaced or energy harvesting circuits whose low efficiencies prevent a frequent and continuous sensing\, even impacting the maximum range of communication. Here\, we discuss a new battery-less and harvester-free remote sensing tag\, namely the subharmonic tag (SubHT)\, leveraging unique nonlinear characteristics to fundamentally break any previous paradigms for passive WSNs. SubHT can sense and transmit information without requiring supplied or harvested DC power. Also\, it transmits the sensed information at a difference frequency from the one of its interrogation signal\, rendering its reader immune from multi-path\, from clutter and from its own self‑interference. Also\, even though SubHT may not require any advanced and expensive manufacturing\, its unique nonlinear response enables extraordinary high sensitivities and dynamic ranges that can even surpass those achieved by the most advanced on-chip sensors. More interestingly\, SubHT can be even configured to operate in a “threshold sensing” mode\, making it able to respond to any interrogation signal only when the sensed parameter has exceeded a remotely reprogrammable threshold\, as well as to memorize any violation in a sensed parameter without requiring any memory components. In this talk\, the first SubHT prototypes for temperature sensing will be showcased. Even more\, we will show how including high quality factor (Q) resonators in a SubHT’s network allows to implement even more functionalities\, such as the long-range identification or tracking of any items or localization and navigation in a GPS denied environment. Yet\, the dynamics exploited by SubHT can also be leveraged to address various needs along radio-frequency (RF) chains. In this regard\, we show how the SubHT’s nonlinear dynamics can be leveraged to build components\, such as parametric filters\, frequency selective limiters and signal to noise enhancers\, that improve the stability of RF frequency synthesizers and instinctually suppress co-site or self-interferes\, paving an unprecedented path towards integrated radios with improved performance and longer battery-life time.
URL:https://ece.northeastern.edu/event/hussein-husseins-phd-dissertation-defense/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230412T150000
DTEND;TZID=America/New_York:20230412T170000
DTSTAMP:20260416T232904
CREATED:20230405T175342Z
LAST-MODIFIED:20230405T175342Z
UID:6250-1681311600-1681318800@ece.northeastern.edu
SUMMARY:IER Open House
DESCRIPTION:Following our success with last year’s Open House\, we are again opening the lab up to visitors! This year\, demos will be held on the 5th floor of ISEC by Yingzi Lin\, Ilya Vidrin\, Alireza Ramezani\, Taskin Padir\, Kris Dorsey\, Rob Platt\, and Hanu Singh. In Richards Hall\, Dagmar Sternad\, CJ Hasson\, and Max Shepherd will be hosting demos as well.\n\n\nThe flyer for this event is attached. Please reach out to Noah (n.smith@northeastern.edu) with any questions! Make sure to register here: https://www.eventbrite.com/e/institute-for-experiential-robotics-open-house-tickets-603775106597
URL:https://ece.northeastern.edu/event/ier-open-house/
LOCATION:532 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230411T120000
DTEND;TZID=America/New_York:20230411T130000
DTSTAMP:20260416T232904
CREATED:20230405T175216Z
LAST-MODIFIED:20230405T175216Z
UID:6248-1681214400-1681218000@ece.northeastern.edu
SUMMARY:Tirthak Patel's Dissertation Defense
DESCRIPTION:“Robust System Software for Quantum Computing” \nCommittee Members: \nProf. Devesh Tiwari (Advisor) \nProf. David Kaeli \nProf. Ningfang Mi \nProf. Gene Cooperman \nProf. Kenneth Brown \nAbstract: \nDespite rapid progress in quantum computing in the last decade\, the limited usability of quantum computers remains a major roadblock toward its wider adoption. Current noisy intermediate-scale quantum (NISQ) computers produce highly erroneous program outputs for quantum-advantage-proven algorithms — that is\, algorithms that are infeasible or orders of magnitude slower on classical supercomputing and high-performance computing (HPC) clusters. Unfortunately\, currently\, quantum computing programmers lack robust system software tools and methods to make meaningful use of erroneous program executions on quantum computers. \nThis lack of capability is the core motivation behind the fundamental question this dissertation poses: “can we build system software tools for programmers to make the quantum program execution and output meaningful on NISQ machines?” This dissertation answers this question in the affirmative— experimentally demonstrating on real-system quantum computers that it is possible to extract near-accurate program output from noisy executions on today’s erroneous quantum computers\, ironically using classical HPC resources and knowledge. This dissertation demonstrates how to achieve this goal without requiring user intervention\, domain knowledge about quantum algorithms\, or additional quantum hardware support. \nUnfortunately\, as this dissertation uncovers\, progressing toward making quantum computers usable is a double-edged sword. In the near future\, only a few entities in the world may have access to powerful quantum computers\, and these quantum computers will be used to solve previously-unsolved large-scale optimization problems\, possibly without an explicit trust model between the service provider and the customer. Therefore\, this dissertation envisions that the solutions to such large-scale optimization problems will be considered sensitive and will need to be protected. This dissertation takes the first few steps toward preparing us for that future by developing a novel method that intelligently obfuscates near-accurate program output and quantum circuit structure to preserve a customer’s privacy under a specified computation model and resource availability. \nThe approaches introduced in this dissertation open up new research avenues for hybrid quantum-classical computing and lower the barrier to entry for quantum computing research for the experimental computer systems and HPC community by open-sourcing multiple novel datasets and software frameworks implemented for real-system quantum computers. \nCandidate Bio: \nTirthak Patel is an incoming Assistant Professor in the Department of Computer Science at Rice University; currently\, a PhD candidate at Northeastern University\, advised by Professor Devesh Tiwari. Tirthak conducts systems-level research at the intersection of quantum computing and high-performance computing (HPC). His research contributions have appeared at rigorously peer-reviewed publication venues including ASPLOS\, Supercomputing (SC)\, HPDC\, HPCA\, and USENIX FAST\, and have been recognized with multiple award distinctions. He has received the ACM-IEEE CS George Michael Memorial HPC Fellowship\, the NSERC Alexander Graham Bell Canada Graduate Scholarship\, and the Northeastern University Outstanding Graduate Student in Research award\, for his research contributions toward making noisy quantum computing systems useful and helping HPC programmers solve computationally challenging problems.
URL:https://ece.northeastern.edu/event/tirthak-patels-dissertation-defense/
LOCATION:442 Dana\, 360 Huntington Ave\, 442 DA\, Boston\, MA\, 02115\, United States
GEO:42.3387508;-71.0923044
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=442 Dana 360 Huntington Ave 442 DA Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 442 DA:geo:-71.0923044,42.3387508
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230404T123000
DTEND;TZID=America/New_York:20230404T133000
DTSTAMP:20260416T232904
CREATED:20230328T174955Z
LAST-MODIFIED:20230328T174955Z
UID:6229-1680611400-1680615000@ece.northeastern.edu
SUMMARY:Cheng Gongye's MS Thesis Defense
DESCRIPTION:“Using Floating-Point Timing Side-Channels to Reverse Engineer Deep Neural Networks” \nCommittee Members:\nProf. Yunsi Fei (Advisor)\nProf. Aidong Ding\nProf. Xiaolin Xu \nAbstract: \nTrained Deep Neural Network (DNN) models have become valuable intellectual property. A new attack surface has emerged for DNNs: model reverse engineering. Several recent attempts have utilized various common side channels. However\, recovering DNN parameters\, weights and biases\, remains a challenge. In this paper\, we present a novel attack that utilizes a floating-point timing side channel to reverse-engineer parameters of multi-layer perceptron (MLP) models in software implementation\, entirely and precisely. To the best of our knowledge\, this is the first work that leverages a floating-point timing side channel for effective DNN model recovery.
URL:https://ece.northeastern.edu/event/cheng-gongyes-ms-thesis-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230403T100000
DTEND;TZID=America/New_York:20230403T110000
DTSTAMP:20260416T232904
CREATED:20230320T205517Z
LAST-MODIFIED:20230320T205517Z
UID:6217-1680516000-1680519600@ece.northeastern.edu
SUMMARY:Jared Miller Ph.D Defense/Proposal Announcement
DESCRIPTION:“Safety Analysis for Nonlinear and Time-Delay Systems using Occupation Measures” \nInternational Village 022 \nCommittee Members:\nProf. Mario Sznaier (Advisor)\nProf. Octavia Camps\nProf. Bahram Shafai\nProf. Eduardo Sontag\nProf. Didier Henrion (LAAS-CNRS) \nAbstract:\nThis research extends an occupation measure framework to analyze the behavior and safety of dynamical systems. A motivating application of trajectory analysis is in peak estimation\, which finds the extreme values of a state function along trajectories. Examples of peak estimation include finding the maximum height of a wave\, voltage on a power line\, speed of a vehicle\, and infected population in an epidemic. Peak estimation can be applied towards safety quantification\, such as by measuring the safety of a trajectory by its distance of closest approach to an unsafe set. \nA finite-dimensional but nonconvex peak estimation problem can be converted into an infinite-dimensional linear program (LP) in measures\, which is in turn bounded by a convergent sequence of semidefinite programs. The LP is posed in terms of an initial\, a terminal\, and an occupational measure\, where the occupation measure contains all possible information about the dynamical systems’ trajectories. This research applies measure-based methods towards safety quantification (e.g. distance estimation\, control effort needed to crash)\, hybrid systems\, bounded-uncertain systems (including for data-driven analysis)\, stochastic systems\, and time-delay systems. The modularity of this measure-based framework allows for multiple problem variations to be applied simultaneously (e.g. distance estimation under time-delays)\, and for optimization models to be synthesized using MATLAB. Solving these optimization problems results in certifiable guarantees on system performance and behavior.
URL:https://ece.northeastern.edu/event/jared-miller-ph-d-defense-proposal-announcement/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230322T180000
DTEND;TZID=America/New_York:20230322T190000
DTSTAMP:20260416T232904
CREATED:20230131T012350Z
LAST-MODIFIED:20230131T012350Z
UID:6072-1679508000-1679511600@ece.northeastern.edu
SUMMARY:PlusOne Information Session
DESCRIPTION:LEARN ABOUT THE PLUSONE ACCELERATED MASTER’S DEGREE PROGRAM \nA master’s degree can provide you an additional level of expertise in an area aligned with your career goals. As a currently enrolled Bachelor of Science (BS) student in the College of Engineering at Northeastern\, you have the opportunity to earn a Master of Science degree (MS) in an accelerated time period with the PlusOne program. Once accepted into the program in an approved PlusOne pathway\, which is a BS and MS PlusOne combination\, you can earn an MS degree with\, in most cases\, just one extra year of study beyond your undergraduate degree program. \nIn this virtual information session\, College of Engineering undergraduate and graduate academic advisors will provide an overview of the PlusOne program to give you the knowledge and next steps to take advantage of the program if you choose. \nWHAT YOU WILL LEARN: \n\nWhat is PlusOne\nBenefits of the program\nEligibility\nCo-op considerations\nFinancial considerations\nSelecting your pathway\nAcademic advising resources\nTimeline to apply\nThe application process\nCourse registration\nTransitioning to graduate school\n\nZoom
URL:https://ece.northeastern.edu/event/plusone-information-session-4/
LOCATION:MA
ORGANIZER;CN="Graduate School of Engineering":MAILTO:coe-gradadmissions@northeastern.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230315T140000
DTEND;TZID=America/New_York:20230315T153000
DTSTAMP:20260416T232904
CREATED:20230315T181507Z
LAST-MODIFIED:20230315T181507Z
UID:6210-1678888800-1678894200@ece.northeastern.edu
SUMMARY:Sadjad Asghari Esfeden's PhD Dissertation Defense
DESCRIPTION:“Spatiotemporal Localization of Object Handover for Human Robot Collaboration” \nCommittee Members: \nProf. Deniz Erdogmus (Advisor) \nProf. Taskin Padir \nProf. Eugene Tunik \nProf. Mathew Yarossi \nAbstract: \nHuman-robot interaction in a physical world like handover of objects requires perception systems to be efficient in localizing the object of interest. We propose an approach to estimate the location of the object with a low-cost RGB camera in a real-time inference for human-robot handover. While handover can take place in a short amount of time\, it is important for a robot to keep track of the object and fill in the gaps of missing detections in the perception module\, especially when the object is partially or completely occluded. A robot needs to proactively detect and track the object since the human decides where and when to transfer the object to the robot in a human to robot object handover.  In order to develop a perception system for robot to be capable of constantly localizing the object and predict its location and time of transfer\, we integrate an object detection algorithm with a tracking framework. The evaluation of this pipeline shows promising results for the goal of localization and tracking of the handover object and can help its location prediction in future.
URL:https://ece.northeastern.edu/event/sadjad-asghari-esfedens-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230303T140000
DTEND;TZID=America/New_York:20230303T153000
DTSTAMP:20260416T232904
CREATED:20230227T195344Z
LAST-MODIFIED:20230227T195344Z
UID:6157-1677852000-1677857400@ece.northeastern.edu
SUMMARY:Kerem Enhos' PhD Proposal
DESCRIPTION:“Software-Defined Inter-medium Visible Light Communication and Underwater Acoustic Networks” \nCommittee Members:\nProf. Tommaso Melodia (Advisor)\nProf. Kaushik Chowdhury\nProf. Stefano Basagni\nDr. Emrecan Demirors \nAbstract:\n“Multi-Domain Operations” paradigm has been receiving significant attention both in military and civilian worlds. To realize this novel paradigm\, it is imperative to establish robust communication links to transfer data between devices operating in multiple domains. However\, as of today\, establishing high data rate\, robust\, secure\, and bi-directional communication links between aerial and underwater assets across the air-water interface is still an open problem. We address these challenges with software-defined visible light networking to establish bi-directional wireless links through the air-water interface. After generating a simulation model for inter-medium communication channel\, we also empirically derived an optimal parameter selection for carrierless amplitude and phase (CAP) modulation. Then\, we design and prototype a software-defined visible light  communication (VLC) modem and conducted extensive experimental evaluation. Apart from inter-medium communication\, software-defined networking can also be leveraged for underwater acoustic communication (UWAC)\, where we designed and assessed coexistence of multi-dimensional chirp spread spectrum (MCSS) with other UWAC schemes. We first evaluated the performance of the proposed communication scheme in a heterogeneous network setting  where it co-exists with a ZP-OFDM communication link\, then in a homogeneous network setting where all links are using MCSS scheme. Finally\, we used  this software-defined networking system to implement a single-input  multiple-output (SIMO) system for UWAC modems that are  deployed in a  distributed manner. Then\, we conduct a thorough experimental evaluation in  ocean environment for various subcarrier bandwidths and constellations  using three distributed receivers.
URL:https://ece.northeastern.edu/event/kerem-enhos-phd-proposal/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230303T100000
DTEND;TZID=America/New_York:20230303T110000
DTSTAMP:20260416T232904
CREATED:20230223T212329Z
LAST-MODIFIED:20230223T212329Z
UID:6152-1677837600-1677841200@ece.northeastern.edu
SUMMARY:Guanying Sun's PhD Dissertation Defense
DESCRIPTION:“Optimizing Reconstruction for Mm-Wave Body Scanner Imaging” \nCommittee: \nProf. Carey Rappaport (Advisor) \nProf. Edwin Marengo \nProf. Jose Martinez-Lorenzo \nAbstract: \nIn the past decades\, due to evolving threats\, passenger screening has become an important secure measure at airport and other secure locations. Numerous passenger screening techniques have been developed by researchers in both academia and industry to detect threats from explosives and weapons. Among these developments\, the multistatic mm-wave radar Advanced Imaging Technology (AIT) system was developed at Northeastern University. A problem with this system is the sidelobes from its physical limitations\, such as the finite aperture extent and the violation of the Nyquist sampling criterion by the sparse array. Therefore\, it is important to suppress the sidelobes so that to improve the quality of the reconstruction image. In this proposal\, we investigate two categories of methods\, one is based on post-processing\, and the other is based on system configuration optimization. In the former category four methods are developed\, while in the latter two methods are proposed. \nIn the first category\, the first method is the phase coherence method which is designed to weight the coherent sum based on the phase diversity of the reconstructed solutions for different transmitters. In this method\, two ways are considered to construct the Phase Coherence Factor (PCF). The first way is to use the information of wrapped phase\, and the second way is to use the information of unwrapped phase\, which is more intuitive than the first way. The second method is the coherence factor related method. Three coherence-factor based methods are analyzed and then incorporated into the imaging procedure of our nearfield millimeter-wave radar security scanning system. The third method is the SNR-dependent coherence factor method\, which takes SNR into consideration when forming the coherence factor. This method can generate better results than the pure coherence-factor based methods by choosing a proper set of parameters. The fourth method is the block-weighting algorithm where the neighbor weight amplifies bright areas and attenuates dark areas\, while the block keeps the influence local. The effectiveness of these methods has been verified with both simulation and measurement data. \nIn the second category\, the first method is optimizing receiver positions via PSF-based multi-objective optimization. Two metrics for measuring image quality of the PSF are proposed and defined as objective functions. The solution-selection metric is introduced to select the desired solution from the numerous Pareto-optimal solutions. Simulation shows that the optimized receiver design generates images with lower sidelobe level than the uniform receiver design. The second method is the dual-frequency radar design\, where a dual frequency\, wideband antenna array is designed by combining a high frequency subarray with a low frequency subarray. The image of the dual frequency array is obtained by multiplying the images of the two subarrays. We analyzed the amplitude of the PSF theoretically and proposed a criterion for the selection of dual frequency array design. The system imaging simulation shows that the grating lobes are significantly reduced for the dual frequency array with fewer radar modules/elements than the conventional array. This design will make the new generation system superior to the conventional scanning system.
URL:https://ece.northeastern.edu/event/guanying-suns-phd-dissertation-defense/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230302T090000
DTEND;TZID=America/New_York:20230302T100000
DTSTAMP:20260416T232904
CREATED:20230223T212222Z
LAST-MODIFIED:20230223T212222Z
UID:6150-1677747600-1677751200@ece.northeastern.edu
SUMMARY:Matthew Schinault's PhD Proposal Review
DESCRIPTION:“Development of A Large-Aperture 160-Element Coherent Hydrophone Array System for Instantaneous Wide Area Ocean Acoustic Sensing” \nAbstract: \nA large aperture coherent hydrophone towed array system comprising of 160 elements and an aperture length of 192 meters has been developed for real-time instantaneous wide-area ocean acoustic remote sensing and monitoring. The design and manufacture of these arrays requires a multidisciplinary approach to achieve acoustic performance capable for detection\, classification\, localization and tracking. Drawing from disciplines such as material science\, electrical engineering\, mechanical engineering\, hydrodynamics\, oceanography\, bioacoustics and signal processing. Due to the cost and complexity of towed array technology\, development of large aperture towed arrays has been limited at the university level. With military\, oil and gas exploration as the chief technology developers and users. The military and commercial focus is narrow and does not allow for scientific study\, resulting in significant gaps in the way we understand ocean acoustics around the globe. Here we model\, design\, fabricate and field test a broadband array for general ocean sensing that is configured to support a wide range of research to include study of marine mammals\, fish shoals\, geophysical processes\, surface or subsea man-made craft\, seismic surveying and the various challenges associated with detection\, classification and localization of underwater sound sources. \nHere\, we present the design process\, beginning with modeling and measurement of piezoelectric material properties. This allows us to perform finite element analysis\, estimating beampatterns and frequency response with a hydrophone electrical model. A pressure to voltage input model of the hydrophone is used to obtain the voltage levels produced to then configure amplification\, gain and filter stages providing a system level transfer function from analog to digital conversion. The array performance with a delay and sum beamformer is estimated for a broad range of frequencies\, with beamforming above half-lambda spacing. The components of the mechanical tow package are modeled to inform array construction estimating vibration and flow noise. A turbulent boundary layer model for flow noise estimation and environmental noise model determines the gains and cutoff frequencies necessary for performance. The comprehensive performance model is compared with a parameter estimation from test data to quantify array performance. \nTowed arrays are subject to environmental extremes\, with time at sea being costly. To increase the reliability\, the array is designed using field replaceable pressure tolerant components including hydrophones\, pre-amplifiers\, power modules\, telemetry and analog to digital conversion units. All components are verified by pressure chamber testing to ensure operation at depth. This large aperture array was able to be made without specialized facilities by utilizing modular interchangeable array interconnects allowing for conventional array populating and oil-filling methods with aperture lengths that are serviceable onboard research vessels. Array design\, fabrication and assembly was performed on-site at Northeastern University in Boston\, Massachusetts. Examples of passive acoustic data from array deployment during a sea trial in the U.S. Northeast coast are presented illustrating array capabilities. \nCommittee: \nProf. Purnima Ratilal Makris (Advisor)\nProf. Marvin Onabajo\nProf. Yongmin Liu\nDr. Alessandra Tesei
URL:https://ece.northeastern.edu/event/matthew-schinaults-phd-proposal-review/
LOCATION:MA
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230227T130000
DTEND;TZID=America/New_York:20230227T150000
DTSTAMP:20260416T232904
CREATED:20230131T200745Z
LAST-MODIFIED:20230131T200745Z
UID:6074-1677502800-1677510000@ece.northeastern.edu
SUMMARY:COE PhD Research Expo
DESCRIPTION:The College of Engineering is excited to announce the fifth annual COE PhD Research Expo\, and we invite all COE PhD students to submit a poster abstract. The expo is an excellent opportunity for your students to highlight their research and gain presentation experience before RISE. \nEvent:   COE PhD Research Expo\nDate:     Monday\, February 27\, 2023\nTime:    1:00pm – 3:00pm\nPlace:    McLeod Suites – Curry Student Center \nThe expo will take place following National Engineer’s Week. \nStudent Abstracts: \nPlease encourage your PhD students to submit poster abstracts by February 10\, 2023. The COE Communications Lab will offer interested students a poster preparation and presentation workshop early February. We will send details of the workshop to students soon. \nFaculty Judges: \nWe are looking for around ten faculty members to serve as judges. If you are available to judge between 1:30pm and 3:00pm on Monday\, February 27th please reach out to Taryn Urbanus (t.urbanus@northeastern.edu) by Friday\, February 17th. \n 
URL:https://ece.northeastern.edu/event/coe-phd-research-expo/
LOCATION:Curry Student Center\, 360 Huntington Ave.\, Boston\, MA\, 02115\, United States
GEO:42.3394629;-71.0885286
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Curry Student Center 360 Huntington Ave. Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave.:geo:-71.0885286,42.3394629
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230227T130000
DTEND;TZID=America/New_York:20230227T140000
DTSTAMP:20260416T232904
CREATED:20230223T212432Z
LAST-MODIFIED:20230223T212432Z
UID:6154-1677502800-1677506400@ece.northeastern.edu
SUMMARY:Yu Yin's PhD Proposal Review
DESCRIPTION:Committee: \nProf. Yun Fu (Advisor) \nProf. Sarah Ostadabbas \nProf. Ming Shao \nAbstract:\nThe community has long enjoyed the benefits of synthesizing data\, as it provides a reliable and controllable source for training machine learning models while reducing the need for data collection from the real world. Human face and body synthesis are especially appealing to research communities\, where model fairness and ethical deployment are critical concerns. However\, generating digit humans that are convincing\, realistic-looking\, identity-preserving\, and high-quality are still challenging in 2D and 3D image synthesis.\nThis dissertation investigates the potential for understanding human behavior by recreating it\, and can be broadly divided into three sections. (1) In Section one\, we explore the 2D image generation models and their interaction with face applications (i.e.\, landmark localization and face recognition tasks). Specifically\, super-resolution (SR) and landmark localization of tiny faces are highly correlated tasks. To this end\, we propose joint frameworks that enable face alignment and SR to benefit from one another\, hence enhancing the performance of both tasks. Moreover\, we demonstrate that face frontalization provides an effective and efficient way for face data augmentation and further improves face recognition performance in extreme pose scenarios. (2) In Section two\, we explore the 3D parametric generation models and how they support human body pose and shape estimation. Advancing technology to monitor our bodies and behavior while sleeping and resting is essential for healthcare. However\, keen challenges arise from our tendency to rest under blankets. To mitigate the negative effects of blanket occlusion\, we use an attention-based restoration module to explicitly reduce the uncertainty of occluded parts by generating uncovered modalities\, which further update the current estimation via a cyclic fashion. (3) In Section three\, we explore the 3D Nerf-based Generative models in generating high-quality images with consistent 3D geometry. We propose a universal method to surgically fine-tune these NeRF-GAN models in order to achieve high-fidelity animation of real subjects only by a single image.
URL:https://ece.northeastern.edu/event/yu-yins-phd-proposal-review/
LOCATION:MA
END:VEVENT
END:VCALENDAR