Speaker:  Dr. Radhika Nagpal, Professor of Robotics, Princeton University

Title:  Taming the Swarm


In nature, thousands of individuals can cooperate to achieve complex goals from local interactions -- from cells that form complex organisms, to social insects like ants and bees, to the complex and mesmerizing motion of fish schools and bird flocks. These systems are fascinating to scientists and engineers alike: even though each individual has limited ability, as a collective they seem to behave as one. What would it take to create our own artificial collectives of the scale and complexity that nature achieves? In this talk I will discuss how biological swarms can inspire new forms of artificial intelligence and physical robotic systems.


Radhika Nagpal is a Professor at Princeton University, jointly in Mechanical Engineering and Computer Science Departments, where she leads the Self-organizing Swarms and Robotics Group (SSR).  Nagpal is a leading researcher in swarm robotics, bio-inspired algorithms, and collective intelligence. Projects from her lab include bio-inspired multi-robot systems such as the Kilobot thousand-robot swarm, Termes collective construction robots, and BlueSwarm underwater robots (Science 2014, Science Robotics 2021) as well as models of biological collective intelligence (Nature Comms. 2022). Nagpal is also known for her Scientific American blog article (“The Awesomest 7 Year Postdoc”, 2013) advocating academic cultural change and she received the Anita Borg Early Career Award (2010) and McDonald Mentoring Award (2015).  In 2017 Nagpal co-founded ROOT Robotics, an educational robotics company now part of iRobot. Nagpal was named by Nature magazine as one of the top ten influential researchers of the year (2014) and was a TED speaker in 2017.

Host: Margo Seltzer, UBC Computer Science


Name: Victor Sanches Portella
Date: July 30
Time: 10 am
Location: ICCS X836
Supervisor: Nick Harvey

Title: Privacy, Experts, and Martingales: An Investigation on the use of Analytical Tools
Abstract: In this thesis, we describe new results on three problems in learning theory and probability theory: private estimation of Gaussian covariance matrices, prediction with experts’ advice, and the expected norm of martingales. Interestingly, in all of them one of the key ingredients is the use of analytical tools in mathematics.

Estimation of the covariance matrix of a Gaussian distribution from samples is a classical problem in statistics that has been thoroughly studied in the literature. Recently researchers proposed the model of differential privacy, a mathematical framework to provide formal guarantees on the amount of sensitive information leaked by algorithms. This compelled researchers to revisit classical problems such as covariance matrix estimation to better understand the limits of statistical estimation under differential privacy. In this thesis we provide tight lower bounds on the accuracy of estimation of Gaussian covariance matrices under the broadest regime of parameters compared to previously known lower bounds.

The framework of prediction with experts’ advice is a theoretical model where a player and an adversary play a game with multiple rounds. At each round, the player selects one of multiple experts whose advice to follow while the adversary decides on the cost of following the advice of each expert. In this thesis, we study a continuous-time model of the experts’ problem and provide several results in this setting—that often translate to the discrete problem—with a focus on anytime strategies, that is, those that do not require knowledge on the length of the game. We describe new anytime algorithms with best-known guarantees against the top-quantile of experts in hindsight. Moreover, we show an anytime strategy in continuous time whose guarantees against independent experts match the guarantees of optimal algorithms in the fixed-time setting.

Finally, motivated by our investigations of the continuous-time experts’ problem, we study the problem of bounding the infinity norm of high-dimensional martingales under a large class of stopping times. We show asymptotically tight upper and lower bounds to the expected norm of range of continuous and discrete time martingales, generalizing results known for one dimensional martingales.


Name: Yi Nian (Jeffrey) Niu

Date and Time: July 25th at 1:30pm-3:00pm

Location: X530

Supervisor: Jiarui Ding

Title: Machine learning approaches for single-cell multiomics data integration and generation


Single-cell multiomics technologies generate paired or multiple measurements of different biological modalities, such as gene expression and chromatin accessibility. However, multiomics technologies are more expensive than their single-modality counterparts, resulting in smaller and fewer available multiomics datasets. Here, we present scPairing, a variational autoencoder model inspired by Contrastive Language-Image Pre-Training, which embeds different modalities from the same single cells onto a common embedding space. We leverage the common embedding space to generate novel multiomics data following bridge integration. Through extensive benchmarking, we show that scPairing constructs an embedding space that fully captures both coarse and fine biological structures. Then, we use scPairing to generate new multiomics data from retina and immune cells. Furthermore, we extend to co-embed three modalities and generate a new trimodal dataset of bone marrow immune cells. Researchers can use these generated multiomics datasets to discover new biological relationships across modalities or confirm existing hypotheses without the need for costly multiomics technologies.


Name: Hedayat Zarkoob

Date: July 23

Time: 2-5pm

Location: ICICS X836

Supervisor’s name: Kevin Leyton-Brown

Title of thesis: AI-Powered Methods for Academic Assessment: Overcoming Scalability Challenges in Large University Classrooms and Conference Review

Abstract: In this thesis, we use various AI techniques to address several scalability challenges in two academic environments: large university classrooms and large peer-review conferences. In large university classrooms, two main challenges that instructors face are grading open-ended assignments and facilitating in-class discussions. To tackle the issue of grading open-ended assignments at scale, we use ideas from mechanism design and graphical models to design practical peer grading systems that provide strong incentives for students to be truthful and that accurately aggregate reported grades. To facilitate in-class discussions, we develop and analyze a new web-based participation tool designed to encourage active participation from students of different demographics. For large peer-reviewed conferences, we propose a novel reviewer-paper matching approach that uses machine learning and mixed-integer programming techniques to preserve the quality of reviews by finding better matches between reviewers and papers and using reviewer resources more efficiently. To demonstrate the effectiveness of the innovations introduced, we evaluate each innovation through analysis on both real and synthetic data, as well as through survey data.


Name: Ruiyu Gou

Date & Time: July 23rd 3 - 4 pm

Location: ICCS 238

Supervisor: Michiel van de Panne

Title: Learning Temporal Action Chunking for Motor Control


Deep reinforcement learning has had significant success at learning motor control tasks. Typically, these policies are fully closed loop or `state indexed', implying a control policy that is queried at every control time step with the current state in order to estimate the best current action corresponding to that state. However, this approach ignores the inherent predictability of many systems, wherein the future states and actions are often quite predictable and can thus be controlled in an open-loop or `time indexed' fashion.

Chunking of action sequences is a well-established mechanism in cognitive system to enhance memory and efficiency during task learning and execution. By modelling actions in temporal chunks, one reduces the computational and perceptual demands required for control. Learning this type of temporal action abstraction remains an under-explored direction.

We present a method that learns a chunk-based state-and-time-indexed policy from any existing state-indexed reinforcement learning policy, with minimal added complexity. We show that with a straightforward multi-layer perception, the chunk-based policy can decrease the required control frequency significantly.  In particular, we show a reduction from 60Hz to 10Hz for the control of a fully 3D humanoid capable of robust and realistic movement across varying terrain. We further propose an adaptive runtime algorithm that can leverage long action chunks while reverting to single-step actions as needed in order to achieve robust behavior.