USC Viterbi Summer Undergraduate Research Experience (SURE)
Computer Science
Project Titles
Click on each project title to find a description of the lab and any preferred background. Many of these labs are interdisciplinary, so please look through projects that are housed in other departments.
We will be adding more labs during the weeks, so please come back to check out any additional labs that have been posted.
Faculty: Quan Nguyen
Project Description: Design and Control of a Highly Dynamic Wheel-legged Robot. A video of the current result can be seen at: https://youtu.be/9vkS0IoGp0s
Related Background: Computer Science, Mechanical Engineering, Electrical Engineering
Prerequisites: Background in Design, Control, Robotics
Faculty: David Barnhart
Project Description: LEAPFROG is a standalone lunar lander prototype vehicle that uses an RC turbine engine and paintball tanks to fly repeatedly to test in simulated lunar landing conditions.
Related Background: EE, MechE, CS, AE
Prerequisites: Some hands on EE experience, GNC knowledge
Faculty: David Barnhart
Project Description: CLING-ERS is a genderless docking system that has gotten approval to be tested onboard the ISS in the summer of 2022.
Related Background: EE, ME, AE, CS
Prerequisites: Some hands on EE experience, RTOS programming (C or Python)
https://www.isi.edu/centers/serc/rendezvous_and_proximity_operations_rpo_research
Faculty: David Barnhart
Project Description: STARFISH is a unique robotic system meant to operate in space and uses a completely soft flexible material and distributed processing to operate very similar to a biological starfish does.
Related Background: EE, CS, AE with Controls background
Prerequisites: Some hands on EE experience, RTOS programming (C or Python), 3D printing experience
https://www.isi.edu/centers/serc/rendezvous_and_proximity_operations_rpo_research
Faculty: Shaama Sharada
Project Description: While machine learning methods are rapidly and successfully being adopted for predicting material and chemical properties, the inverse problem of predicting novel materials with desired properties is still in its infancy. We are implementing deep learning frameworks to address this problem and to explore the chemical space based on electronic and steric properties of ligands in transition metal complex catalysts.
Related Background: Introductory machine learning course desirable
Prerequisites: Python/programming
Faculty: Jesse Thomason
Project Description: An agent interpreting natural language instructions in a real or simulated world needs to identify salient objects in the world to which that language refers. Many methods for benchmarks like ALFRED [https://askforalfred.com/] use off-the-shelf object detectors like Faster R-CNN, minimally fine-tuning them for objects in the 3D environment. We hypothesize that understanding what kind of object an agent is looking for can influence the accuracy of object detections. For example, a detector trained specifically to recognize "pickupable" objects will more likely detect small objects like forks and spoons, one for "openable" objects cabinet and drawer faces, and one for "closable" objects cabinet and drawer interiors. The project will involve using action types as conditioning information for object detectors in a language-guided task completion benchmark. The student will become familiar with the ALFRED benchmark and with one or more state of the art models that tackle the challenge. We will aim to achieve a new state of the art by adding action-conditioned visual object recognition.
Related Background: Should be comfortable programing in python. Familiarity with pytorch would be a plus.
Prerequisites: CSCI 360 or CSCI 467
Faculty: Muhao Chen
Project Description: Human languages evolve to communicate about events happening in the real world. Therefore, understanding events plays a critical role in natural language understanding (NLU). A key challenge to this mission lies in the fact that events are not just simple, standalone predicates. Rather, they are often described at different granularities, temporally form event processes, and are directed by specific central goals in a context. Our research in this line helps the machine understand events described in natural language. This includes the understanding of how events are connected, form processes or structure complices, and the recognition of typical properties of events (e.g., space, time, salience, essentiality, implicitness, memberships, etc.). This project is to target publication(s) in top-tier ML/NLP conferences or journals.
Related Background: solid background in NLP or machine learning (deep learning), proficiency in PyTorch and Hugging Face
Prerequisites: solid background in NLP or machine learning (deep learning)
Faculty: Gale Lucas
Project Description: The Affective Computing Group at USC’s Institute for Creative Technologies (ICT) is seeking an intern to assist our research in automated negotiation systems that exhibit appropriate emotional and strategical awareness, while utilizing realistic modes of communication such as natural language. Interns will be involved in designing a practically-inspired dialogue systems that engages in effective negotiations with humans. This is a highly interdisciplinary project. Such negotiation systems find a number of practical applications from pedagogy to conversational AI. The precise problem can be decided by incorporating the interests of the interns but potential problems include: 1) manipulating the emotional or strategical behaviors of the negotiation system and conducting crowdsourcing experiments to understand how these manipulations impact negotiation outcomes, 2) using insights from prior affective computing or psychological studies to build neural text generation models for negotiation dialogue systems, 3) improving response quality in terms of consistency and coherency, 4) opponent modelling approaches based on chat-based conversations. Our summer interns frequently publish their work at a wide variety of venues. Here are a couple of successful intern projects published at NAACL 2021 (https://aclanthology.org/2021.naacl-main.254.pdf) and ACII 2021 (https://arxiv.org/pdf/2107.13165.pdf). We expect future work to build up on this progress.
Related Background: refer Computer Science but open to diverse backgrounds if strong skills in pre-requisites
Prerequisites: Strong coding skills (Python) and preferred prior experience/interest in Web development and machine learning
Faculty: Muhao Chen
Project Description: Knowledge acquisition (e.g., relation extraction, entity and event typing) faces challenges including extreme label spaces, few-shot/zero-shot predictions and out-of-domain prediction. To this end, we study methods for leveraging indirect supervision signals from auxiliary tasks (e.g., natural language inference, text summarization, etx.) to foster robust and generalizable inference for knowledge acquisition. In the same context, we study methods for generating semantically rich label representations based on either gloss knowledge or structural knowledge from a well-populated lexical knowledge base, in order to better support learning with limited labels. This project is to target publication(s) in top-tier ML/NLP conferences or journals.
Related Background: proficiency in pytorch and huggingface
Prerequisites: solid background in NLP or machine learning (deep learning), proficiency in PyTorch and Hugging Face
Faculty: Muhao Chen
Project Description: Knowledge graphs (KGs) provide both open-world and domain-specific knowledge representations that are integral to many AI systems. However, constructing KGs is usually very costly and requires extensive effort. A widely attempted solution is to learn knowledge acquisition models that automatically induce structured knowledge from unstructured text. However, such models developed through data-driven machine learning are usually fragile to noise in learning resources, and may fall short of providing reliable inference on large, heterogeneous real-world data. We are developping a general meta-learning framework that seeks to systematically improve the robustness of learning and inference for data-driven knowledge acquisition models. We seek to solve several key problems to accomplish the goal: (i) How to identify incorrect training labels and prevents overfitting on noisy labels; (ii) how do detect invalid input instances in inference (e.g., out-of-distribution ones) and provide abstention-awareness; (iii) automated constraint learning that strengthens model inference with global consistency; (iv) how to automatically augment training signals of the knowledge acquistion model or the backbone language model. This project is to target publication(s) in top-tier ML/NLP conferences or journals.
Related Background: solid background in NLP or machine learning (deep learning), proficiency in PyTorch and Hugging Face
Prerequisites: solid background in NLP or machine learning (deep learning)
Faculty: Heather Culbertson
Project Description: This project focuses on the design, building, and control of haptic devices for virtual reality. Current VR systems lack any touch feedback, providing only visual and auditory information to the user. However, touch is a critical component for our interactions with the physical world and with other people. This research will investigate how we use our sense of touch to communicate with the physical world and use this knowledge to design haptic devices and rendering systems that allow users to interact with and communicate through the virtual world. To accomplish this, the project will integrate electronics, mechanical design, programming, and human perception to build and program a device to display artificial touch sensations to a user with the goal of creating a natural and realistic interaction.
Related Background: Background in computer science, electrical engineering, mechanical engineering, or related majors. Experience with circuits and mechanical design a plus, but not required.
Prerequisites: Programming experience (C++ preferred)
Faculty: Meisam Razaviyayn
Project Description: The goal of this project is to train neural networks using measures of performance other than accuracy in the presence of content-shifts. This is particularly important in applications such as classification of hateful/misinformation posts on social media platforms. In this application, the number of positive samples (sample posts containing misinformation) is small compared to the entire number of samples. Hence non-decomposable measures of performance, such as AUROC or accuracy at the top, are needed for auto-enforcement of the integrity-related policies on social media platforms. However, these measures of performance are vulnerable to content shifts. Our goal is to develop scalable algorithms for training neural networks based on measures of performance related to accuracy at the top. Furthermore, the resulting model needs to be robust against content-shifts. This is because the topic of misinformation changes over social media platforms over time.
Related Background: Basic knowledge of machine learning and neural networks. Being familiar with PyTorch and TensorFlow.
Prerequisites: Machine Learning, PyTorch, TensorFlow
Faculty: Vatsal Sharan
Project Description: The student will explore foundational questions regarding machine learning, using a combination of systematic experiments and theoretical analysis. Both students interested in performing systematic experiments to tease out phenomenon in practice, and those interested in using theoretical tools to prove novel guarantees are welcome. The exact questions and their scope is broad, but the following are some potential options: 1. Understanding deep learning: One particular question of interest here is to understand why neural networks generalize despite having the capacity to overfit. We've been exploring what role the data itself has to play in this mystery, exploring connections to the amazing self-supervised learning capability of neural networks in the process. 2. Data augmentation and amplification: In some recent work ("Sample Amplification"), we showed that it is often possible to generate new samples from a distribution without even learning it. We will explore how this ties into various data augmentation techniques, and develop new frameworks to increase dataset size. 3. Fairness and robustness: We will explore how to train models which are robust in many ways, such as to changes in the data distribution, or to do well on minority sub-populations (and not just on average over the entire data). 4. Computational-statistical tradeoffs: This is a more theoretical direction, to understand when computational efficiency might be at odds with statistical requirements (the data needed to learn). Recent work has opened by much uncharted territory, particularly with respect to the role of memory in learning, which we will explore.
Related Background: Some basic understanding of machine learning would be very helpful for certain projects.
Prerequisites: Familiarity with probability, linear algebra, calculus, and analysis of algorithms.