YSP Alumni

Alumni Directory

2020 YSP Program Coordinators: Salima Amiji and Natasha Zaarour


YSP 2020 – Final Video


Participating Labs

LabYSP StudentsTitleAbstract
Faculty:
Elhamifar, Ehsan

Mentors:
Zijia Lu, Yuhan Shen
Alejandro Gardinier,
Amina Naidjate
Summarization using Artificial Intelligence

Poster
Humans have remarkable ability to summarize visual and textual data, e.g., describe the content of a long movie or a news article concisely with few sentences. In this project, students will explore automatic methods based on artificial intelligence to summarize large collections of data such as videos and text. This involves learning and experimenting with combinations of content detection and recognition with summarization. Students will learn to build an automatic algorithm that takes as input a long movie and produces a trailer of the desired length specified by the user. Students will also apply the same method to summarization of news articles.
Faculty:
Fei, Yunsi

Mentor:
Cheng Gongye
Faraz Iqbal,
Jessica Liao
Differential Analysis of Power Data Using Matlab

Poster
To secure information infrastructures like internet, a security engine provides all sorts of cryptographic operations, including encryption, description, signing, authentication, etc. These cryptographic algorithms are open-source and standardized by organizations like NIST, and the only secret of their implementation is their key. However, the key can be reverse engineered by analyzing the power traces of cryptographic execution. In this project, the participating students are provided with data traces obtained in the lab. They will use the Matlab tool to visualize the power traces, and apply some simple statistical tool to retrieve the secret key.
Faculty:
Hashmi, Sara

Mentors:
Lisa Pepdjonovic
Gyuchan Lee,
Alice Zhang
Exploring active particle flows using agent-based-modeling

Poster
Active particles, from large to small, like cars, birds, fish, fire ants, and bacteria, can spontaneously form patterns and exhibit collective dynamics that are unexpected based on the movement of a single agent. Examples of these patterns include traffic jams, swarming behaviors, vertical column formations of ants, and biofilms. Understanding how these patterns emerge from collective behavior can help us mitigate their negative effects, particularly in the case of traffic jams on the large end of the size-scale, and bacterial biofilm formation, which can lead to infectious diseases. In this project, we will use agent-based-modeling simulations in NetLogo software to explore questions about active particles in flow. We will begin with simple rules for the agents to follow while moving or flowing in a simple system, like bacteria flowing through a straight, narrow channel. We will then move on to more complicated systems, including branching flows with particles of different sizes and shapes, to investigate the emergence of collective behaviors.
Faculty:
Kane, Michael

Mentor:
Kris Govertsen,
Jonathan Cohen
Alexia Marriott,
Franklin Ollivierre III
Powering Remote Islands with Tidal Energy

Poster
The ebb and flow of the tides hides a tremendous amount of renewable energy that could be accessible to many remote communities. Unlike other renewable energy sources, tidal energy can be accurately predicted out centuries, based on the cycles on the moon. Turbines and generators are now being developed to harvest this energy. However, a challenge remains: the harvested energy must be stored to fill in the daily and monthly fluctuations. Battery energy storage costs are rapidly decreasing and becoming more affordable. A variety of battery chemistries on the market each have their niche application, so energy storage on a remote island must include a mix of chemistries. This project will apply optimization and parallel computing to explore how different battery chemistries can be combined to meet the energy demands of remote islands.
Faculty:
Lin, Xue "Shelley"

Mentors:
Chenglong Lin, Mengshu Sun
Kenneth Kalin,
Linda Kebichi
Efficient Deep Learning System Design

Poster
Since the Convolutional Neural Networks (CNNs) were exemplified by the performance improvements obtained by AlexNet in 2012, neural network-based computer vision has improved and progressed, achieving superhuman performance that can detect, classify and segment within a complicated image. However, CNN models have very large model size, making them very challenging to be implemented on edge devices for on-device inference. This project will focus on accelerating deep neural networks’ inference execution on edge devices leveraging model compression and FPGA (Field Programmable Gate Arrays) design techniques. We will design new model compression algorithms that can facilitate FPGA implementation and optimization.
Faculty:
Milane, Lara
/Amiji, Mansoor
Rania Alshawabkeh,
Felix Xu
Cancer Biology and Therapeutics Virtual Laboratory

Poster
Students will work in the virtual laboratory to design and virtually evaluate a new cancer therapy. Students will evaluate their new therapy at the formulation bench, ex vivo bench, and in vitro bench. They will explore how to overcome multidrug resistance through design optimization.
Faculty:
Oakes, Jessica

Mentor:
Jacqueline Matz
Katherine Klosterman,
Oliver Trejo
Tissue Micro-Structure Evaluation Following Chronic Cigarette Exposure

Poster
The Integrated Cardiovascular and Pulmonary Team (ICAP, led by Professor Bellini and Professor Oakes in the Department of Bioengineering at Northeastern University) is interested in the long-term health consequences following exposure to cigarette smoke and electronic cigarettes (e-cigs) aerosols. The goals of these projects are to quantify cardiopulmonary dysfunction that occurs in mice as a result of chronic exposure. At the conclusion of the study, we collect and analyze histological images to assess airspace size, collagen/elastin ratios, etc. This project will be focused on analyzing these images to compare micro-structural changes following chronic exposure.
Faculty:
Shrivastava, Aatmesh

Mentors:
Nikita Mirchandani,
Yuqing Zhang
Aribah Baig,
Alessandro Barbiellini-Amidei
Analog Computing based Feature Extraction Technique to Detect Seizures using EEG Signals

Poster
Advances of machine learning algorithms have led to improvements of seizure detection capabilities in monitoring systems based on electroencephalography (EEG). Seizure detection hardware requires accurate feature extraction, which is conventionally done in the digital domain by extracting power in different EEG frequency bands over a particular time window. This project will explore methods to presents an analog counterpart to digital feature extraction. We expect to a significant gain with energy and area utilization using analog computing.
Faculty:
Wang, Lu

Mentor:
Shuyang Cao
Jason Liang,
Stephanie Martinez
Question Generation for Text Summarization with Human-in-the-loop

Poster
With the fast growth of information available on the Internet, there exists an imperative and compelling research mandate for summarization systems that can understand and aggregate information drawn from lengthy and massive amounts of documents in an automatic manner, as well as allow users to quickly comprehend the general subject and then identify topics of interest to drill down. Our goal is to design algorithms that generate question “pivots”, to expose document structure and allow users to choose what they want to know more about after reading an abstract. We will use both human annotation and automatic methods to generate both the questions and their corresponding answers by summarizing the input documents.