- talks - lists.cs.princeton.edu

Sunnie Kim will present herFPO "Advancing Responsible AI with Human-Centered Evaluation" on Wednesday, May 7, 2025 at 1:30 PM in CS 302 and Zoom.
by gradinfo＠cs.princeton.edu 02 May '25

02 May '25

Sunnie Kim will present herFPO "Advancing Responsible AI with Human-Centered Evaluation" on Wednesday, May 7, 2025 at 1:30 PM in CS 302 and Zoom. Location: Zoom link: https://princeton.zoom.us/my/sunniesuhyoung The members of Sunnies committee are as follows: Examiners: Olga Russakovsky (Adviser), Andrés Monroy-Hernández, Q. Vera Liao (Microsoft) Readers: Parastoo Abtahi, Jennifer Wortman Vaughan (Microsoft) A copy of her thesis is available upon request. Please email gradinfo(a)cs.princeton.edu if you would like a copy of the thesis. Everyone is invited to attend her talk. Abstract follows below: As Artificial Intelligence (AI) technologies are increasingly transforming how we live, work, and communicate, AI evaluation must take a human-centered approach to realistically reflect real-world performance and impact. In this thesis, I present research that advances human-centered evaluation, and subsequently responsible development and deployment of AI, by integrating knowledge and methods from AI and HumanComputer Interaction (HCI). First, using explainable AI as an example, I discuss the challenges and necessity of human (as opposed to automatic) evaluation and present human evaluation studies that identify new issues and opportunities in explainable AI research. Next, I bring context into the picture and present contextualized studies with real users that provide a holistic and nuanced understanding of users explainability needs and trust in AI. Then, I illustrate the importance of evaluating human-AI interaction through studies that examine how users trust, reliance, and other perceptions and behaviors are shaped by different features of AI technologies. In closing, I discuss the implications of these work, provide recommendations for human-centered evaluation, and describe future directions for responsible AI.

1 0

Dexin Zhang will present his General Exam "Modular Verification and Repair for Networks" on Monday, May 5, 2025 at 10am in CS 402.
by Gradinfo 02 May '25

02 May '25

Dexin Zhang will present his General Exam "Modular Verification and Repair for Networks" on Monday, May 5, 2025 at 10am in CS 402. Committee: Aarti Gupta (adviser), David Walker, Mae Milano Abstract: Large industrial networks connect millions of servers with hundreds of thousands of network devices. To improve these systems’ reliability, researchers have turned to modular verification techniques to break these large networks into smaller pieces to analyze independently. Unfortunately, in doing so, researchers face tradeoffs. More expressive frameworks can verify a wider range of properties, but often place a greater burden on the user to describe network behavior at the interfaces between components. We present CB-graphs, a new framework for modular network control plane verification that navigates these tradeoffs using graph-based abstractions, allowing it to verify a rich set of properties while simplifying the interfaces a user must supply. Users construct CB-graphs that describe which network devices converge before others. Interfaces between nodes in these graphs may contain more or less information, depending on what is convenient and needed to prove the desired properties. We develop a theory of CB-graph-based verification and prove it sound via a formalization in the Lean theorem prover. We exploit our framework to develop a new, scalable method for local repair of network configurations. We implement our verification and repair algorithms in a tool called CB-Ver, and evaluate its performance on a range of benchmarks.

1 0

Meenal Parakh will present her General Exam "AnyBody: A Benchmark Suite for Cross-Embodiment Manipulation" on Tuesday, May 6, 2025 at 4:30 PM in FC 202.
by CS Grad Department 02 May '25

02 May '25

Meenal Parakh will present her General Exam "AnyBody: A Benchmark Suite for Cross-Embodiment Manipulation" on Tuesday, May 6, 2025 at 4:30 PM in FC 202. Committee Members: Jia Deng (advisor), Szymon Rusinkiewicz, Adam Finkelstein Abstract: Generalizing control policies to novel embodiments remains a fundamental challenge in enabling scalable and transferable learning in robotics. While prior works have explored this in locomotion, a systematic study in the context of manipulation tasks remains limited, partly due to the lack of standardized benchmarks. In this paper, we introduce a benchmark for learning cross-embodiment manipulation, focusing on two foundational tasks, reach and push, across a diverse range of morphologies. The benchmark is designed to test generalization along three axes: interpolation (testing performance within a robot category, sharing the same link structure), extrapolation (testing on a robot with a different link structure), and composition (testing on combinations of link structure). We primarily test RL agents trained under multitask objectives and evaluated for their multi-task performance and zero-shot transfer. Our study aims to answer whether morphology-randomized training can outperform single-embodiment baselines, whether zero-shot generalization to unseen morphologies is feasible, and how consistently these patterns hold across different generalization regimes. The results highlight the current limitations of multi-embodiment learning and provide insights into how architectural and training design choices influence policy generalization. Reading List: https://docs.google.com/document/d/1Ey-U4ydkMTrTO6Qdr9FmycVom5daALctNbeyRjw… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

2 1

Xingyu Zhu will present his General Exam "On the Power of Context-Enhanced Learning in LLMs" on Monday, May 5, 2025 at 1:00 PM in FC 106 and via zoom.
by CS Grad Department 01 May '25

01 May '25

Xingyu Zhu will present his General Exam "On the Power of Context-Enhanced Learning in LLMs" on Monday, May 5, 2025 at 1:00 PM in FC 106 and via zoom. Zoom link: https://princeton.zoom.us/my/jupiter Committee Members: Sanjeev Arora (advisor), Danqi Chen, Chi Jin Abstract: We formalize a new concept for LLMs, context-enhanced learning. It involves standard gradient-based learning on text except that the context is enhanced with additional data on which no auto-regressive gradients are computed. This setting is a gradient-based analog of usual in-context learning (ICL) and appears in some recent works. Using a multi-step reasoning task, we prove in a simplified setting that context-enhanced learning can be exponentially more sample-efficient than standard learning when the model is capable of ICL. At a mechanistic level, we find that the benefit of context-enhancement arises from a more accurate gradient learning signal. We also experimentally demonstrate that it appears hard to detect or recover learning materials that were used in the context during training. This may have implications for data security as well as copyright. Reading List: https://docs.google.com/document/d/1_SYs1g8A9cSF__u9dRffQ30ZYAArjijfiyW-IwV… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Amrit Daswaney will present the General Exam "Fair and Efficient Allocation of Indivisible Goods via Pseudomarkets with Price Floors" on May 2nd, 10 am - 12 pm. Friend Center 005.
by Gradinfo 01 May '25

01 May '25

Amrit Daswaney will present the General Exam "Fair and Efficient Allocation of Indivisible Goods via Pseudomarkets with Price Floors" on May 2nd, 10 am - 12 pm. Friend Center 005. Committee: Mark Braverman (adviser), Matt Weinberg, Fedor Sandomirskiy. Abstract: The combinatorial assignment problem involves allocating m indivisible goods to n agents without any monetary transfers such that the resulting allocation is Pareto efficient and each agent is (approximately) envy-free. This problem models many practical settings of interest: from assigning courses to students to allocating truckloads of food to food banks. A classical route, pioneered by Varian (1974) and extended by Hylland and Zeckhauser (1979), is to search for a Competitive Equilibrium from Equal Incomes (CEEI): endow every agent with one unit of artificial currency and find prices so that when each agent buys their favourite affordable bundle, the market clears (i.e., supply equals demand for all goods). With indivisible goods, however, a CEEI need not exist: in a two-agent, one-good instance any price makes either both agents or neither demand the good. Budish (2011) restores existence by introducing an arbitrarily small budget inequality between agents. The rationale behind introducing this budget inequality is to use these small differences to break ties between otherwise identical agents. However, our first result shows they can legitimize highly unequal outcomes, allowing agents with negligibly higher budgets to potentially leverage this advantage across multiple goods. To mitigate this, we introduce price floors and define a CEEI-with-price-floors mechanism. Our second result proves that this concept is equivalent to a discrete analogue of Varian (1974)'s coalition fairness: in the replica economy, no coalition can seize the bundle of another coalition with the same total endowment and make all of its members better off. The equivalence echoes the Debreu–Scarf core-equivalence theorem (1963). Crucially, whenever a CEEI-with-floors can be proved to exist for a structured domain (e.g., additive values, bounded numbers of agents), an EFX allocation exists for that very domain, providing a concrete pathway toward resolving this long standing open problem in fair division. Our third result establishes the existence of CEEI-with-floors when (i) n = 2 agents or (ii) n agents have identical valuations. If time permits, we will also discuss alternative combinatorial and randomized techniques that have been applied to the problem.

1 0

Adithya Bhaskar will present his General Exam "Effective mechanistic interpretability via pruning" on Monday, May 5, 2025 at 3:00 PM in FC 008 and via zoom.
by CS Grad Department 30 Apr '25

30 Apr '25

Adithya Bhaskar will present his General Exam "Effective mechanistic interpretability via pruning" on Monday, May 5, 2025 at 3:00 PM in FC 008 and via zoom. Zoom link: https://princeton.zoom.us/j/98937358810 Committee Members: Danqi Chen (advisor), Sanjeev Arora, Benjamin Eysenbach Abstract: Mechanistic interpretability strives to explain language models bottom-up by understanding their components. In this talk, I will describe how pruning—removing less important parameters or connections in a model—can enable a new approach to interpretability. I will introduce a curious phenomenon as a case study: fine-tuning a language model with different random seeds can lead to similar in-domain performance but exhibit distinct patterns of generalization. Pruning the model reveals that even a single model contains many subnetworks that match its in-domain performance, but generalize differently. By connecting this result to a “grokking”-like phenomenon observed for out-of-domain generalization, I will develop a theory of why these subnetworks exist in the model, and what role they fulfill. Zooming out of this example, we will apply pruning to develop an efficient, accurate, and scalable tool to automatically discover transformer circuits—basic units of mechanistic interpretability. We develop Edge Pruning, a method that focuses on pruning edges rather than the parameters of a language model. Edge Pruning discovers circuits substantially more faithful to the model as measured by KL divergence while being more efficient and scaling to 100x larger models than prior approaches. We will see via an example how this tool enables future work to understand phenomena that only emerge in large (1B+ parameters) models, such as in-context learning. Our results imply that pruning holds great promise for interpretability—both as a paradigm and as a standalone method. Reading List: https://docs.google.com/document/d/1VJ5busQNlb-er_yt7-ae34x2UC0KXZh2irvR0vi… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Stamatis Alexandropoulos will present his General Exam "Princeton365: A Diverse Dataset with Accurate Camera Pose" on Monday, May 5, 2025 at 2:00 PM in FC 005 and via zoom.
by CS Grad Department 30 Apr '25

30 Apr '25

Stamatis Alexandropoulos will present his General Exam " Princeton365: A Diverse Dataset with Accurate Camera Pose " on Monday, May 5, 2025 at 2:00 PM in FC 005 and via zoom. Zoom link: https://princeton.zoom.us/my/stamalex Committee Members: Jia Deng (advisor), Felix Heide , Adam Finkelstein Abstract: Accurately evaluating SLAM systems often comes at the cost of limited data diversity, while diverse datasets usually lack precise camera pose. In this talk, I will introduce Princeton365, a new dataset designed to bridge this gap. It includes 365 videos captured using a 360° camera and calibration boards to provide accurate ground truth camera poses. The dataset features a wide range of scenarios—including indoor, outdoor, and object scanning scenes—and offers synchronized monocular and stereo RGB video, along with IMU data. I’ll also present a new scale-aware evaluation metric for SLAM based on optical flow induced by pose estimation errors, which enables more consistent comparison across different scenes than traditional metrics like Average Trajectory Error. Finally, I’ll introduce a novel benchmark for view synthesis that tackles challenging cases not handled by current datasets, such as non-Lambertian surfaces and full 360° camera trajectories. Reading List: https://docs.google.com/document/d/1iqBGxihNyFdlsjk5eCaNgA8EV0hOObt1Z_CN9JG… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Erich Liang will present his General Exam "ChangeIn: A Benchmark for Self-Calibration of Dynamic Intrinsics of Video Cameras" on Monday, May 5, 2025 at 10:30 AM in FC 202 and via zoom.
by CS Grad Department 30 Apr '25

30 Apr '25

Erich Liang will present his General Exam "ChangeIn: A Benchmark for Self-Calibration of Dynamic Intrinsics of Video Cameras" on Monday, May 5, 2025 at 10:30 AM in FC 202 and via zoom. Zoom link: https://princeton.zoom.us/j/8238970644 Committee Members: Jia Deng (advisor), Szymon Rusinkiewicz , Ellen Zhong Abstract: Accurately tracking camera intrinsics is crucial for achieving 3D understanding from 2D video. However, most 3D algorithms assume that camera intrinsics stay constant throughout a video, which is often not true for many real-world in-the-wild videos. A major obstacle in this field is a lack of dynamic camera intrinsics benchmarks--existing benchmarks typically offer limited diversity in scene content and intrinsics variation, and none provide per-frame intrinsic changes for consecutive video frames. In this paper, we present ChangeIn, a real-world benchmark that provides per-frame ground truth intrinsics annotations for videos with dynamic intrinsics. Compared to prior benchmarks, ChangeIn captures a wider range of intrinsic variations and scene diversity, featuring 143K annotated frames from 386 high-resolution indoor and outdoor videos with dynamic camera intrinsics. To ensure accurate per-frame intrinsics, we build a comprehensive look-up table of calibration experiments and extend the Kalibr toolbox to improve its accuracy and robustness. Using our benchmark, we evaluate existing baseline methods for predicting camera intrinsics and find that most struggle to achieve accurate predictions on videos with dynamic intrinsics. Reading List: https://docs.google.com/document/d/1pu3P9hWcthJRMvBSssPqWLYiq3QZCVfa4p_v50J… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Kathryn Wantlin will present her General Exam "Self-Supervised Multi-Task Exploration and RL" on Friday, May 2, 2025 at 3:00 PM in CS 401.
by CS Grad Department 29 Apr '25

29 Apr '25

Kathryn Wantlin will present her General Exam "Self-Supervised Multi-Task Exploration and RL" on Friday, May 2, 2025 at 3:00 PM in CS 401. Committee Members: Ryan Adams (advisor), Ben Eysenbach, Olga Russakovsky Abstract: Learning from demonstrations to imitate a single behavior in a given environment can require hundreds or thousands of expert (human) demonstrations and learning a distribution of tasks in an environment is even more prohibitively inefficient. This work presents a framework for achieving zero-shot generalization capabilities in a given environment without access to expert data. We replace the imitation learning training phase with a self-supervised exploration phase, training a goal-conditioned critic, policy and variational posterior for inferring goals, and at test time, are able to zero-shot infer a goal from a demonstration and then unroll the associated goal-conditioned policy to imitate the demonstration’s behavior with low regret. We make the key assumption that many behaviors are likely goal-reaching, an inductive bias that inferring an agent’s goal provides a good bias-variance tradeoff compared to inferring their reward function. During the self-supervised exploration phase, I learn temporally contrastive features and utilize several methods for automatic goal selection, comparing these results against oracle goal sampling. Our pipeline is compared against related methods for zero-shot imitation, showing competitive performance against goal-conditioned behavioral cloning across our benchmarked environments. Reading List: https://docs.google.com/document/d/1zhhjPl5Ei1f_awCVUBwduVqi6F1AHZ6iMiJdsiT… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Han Xu will present his General Exam " Relational Netkat" on May 2, 2025 at 10am in CS 402 and Zoom.
by Gradinfo 29 Apr '25

29 Apr '25

Han Xu will present his General Exam " Relational Netkat" on May 2, 2025 at 10am in CS 402 and Zoom. Committee: David Walker (adviser), Zachary Kincaid, Maria Apostolaki Zoom link: [ https://princeton.zoom.us/j/8011315368 | https://princeton.zoom.us/j/8011315368 ] Abstract: Modern networks are complex, dynamic, and large-scale, making network verification a critical challenge. Relational NetKAT extends traditional NetKAT by providing an expressive and efficient framework for reasoning about network behavior under path changes, forwarding table modifications, and network topology updates. It enables all-path verification, ensuring that network-wide invariants hold across diverse changes such as load balancing and routing updates. Compared to sample-based verification of Rela, which fails to cover all possible paths, Relational NetKAT provides a mathematically sound and scalable approach to checking network correctness. Reading List: Textbook: * Types and programming languages. MIT press, 2002. * Handbook of Model Checking Ch 7.8. Paper: Netkat, Kleene algebra and automata theory * Xieyang Xu, Yifei Yuan, Zachary Kincaid, Arvind Krishnamurthy, Ratul Mahajan, David Walker, and Ennan Zhai. 2024. Relational Network Verification. * Carolyn Jane Anderson, Nate Foster, Arjun Guha, Jean-Baptiste Jeannin, Dexter Kozen, Cole Schlesinger, and David Walker. 2014. NetKAT: semantic foundations for networks. * Nate Foster, Dexter Kozen, Mae Milano, Alexandra Silva, and Laure Thompson. 2015. A Coalgebraic Decision Procedure for NetKAT. * Damien Pous. 2015. Symbolic Algorithms for Language Equivalence and Kleene Algebra with Tests. BDD and Model Checking * Bryant Randy, 1992. Symbolic Boolean manipulation with ordered binary-decision diagrams * Kathi Fisler and Moshe Y. Vardi. 2002. Bisimulation Minimization and Symbolic Model Checking. Network Verification * H. Yang and S. S. Lam, "Real-time verification of network properties using Atomic Predicates," 2013 * Peyman Kazemian, George Varghese, and Nick McKeown. 2012. Header space analysis: static checking for networks. * Ahmed Khurshid, Wenxuan Zhou, Matthew Caesar, and P. Brighten Godfrey. 2012. Veriflow: verifying network-wide invariants in real time.

1 0