- talks - lists.cs.princeton.edu

Sadhika Malladi will present her FPO "Deep Learning Theory in the Age of Generative AI" on Thursday, November 20, 2025 at 11:00 AM in CS 401.
by CS Grad Department 17 Nov '25

17 Nov '25

Sadhika Malladi will present her FPO "Deep Learning Theory in the Age of Generative AI" on Thursday, November 20, 2025 at 11:00 AM in CS 401. The members of Sadhika’s committee are as follows: Examiners: Sanjeev Arora (Adviser), Danqi Chen, Chi Jin Readers: Tri Dao, Zhiyuan Li (TTIC) 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: This thesis develops theoretical frameworks for understanding modern deep learning in realistic training regimes. As model architectures, optimization methods, and data pipelines continue to evolve, much of the existing theory no longer describes how large-scale systems behave in practice. Exhaustive empirical studies, once central to understanding these systems, have become infeasible at current computational scales. The goal of this work is to bridge that gap by constructing mathematical analyses that remain faithful to real training scenarios while offering prescriptive guidance for practice. The first part establishes stochastic differential equation (SDE) approximations that accurately model stochastic optimization at finite learning rates. These analyses clarify when SDE-based descriptions are valid and yield general scaling rules for efficient large-batch training. The second part examines the fine-tuning of pre-trained language models, formalizing it as a small, structured modification to an existing representation. This perspective explains the effectiveness of parameter-efficient adaptation methods and motivates a memory-efficient fine-tuning algorithm derived from the theoretical analysis. The third part studies preference-based alignment, identifying why commonly used objectives fail to fit preference data and introducing diagnostic measures that link training objectives to observed model behavior. Together, these results demonstrate how mathematically grounded analysis can extend beyond idealized assumptions to capture the dynamics of contemporary machine learning systems. The thesis argues that such theory can play a prescriptive and interpretive role—guiding large-scale training, diagnosing failure modes, and providing a stable foundation for understanding as empirical practices continue to evolve.

2 1

Abhishek Joshiwill present his MSE talk "Procedural Generation of Articulated Simulation-Ready Assets" TODAY (Monday, November 17, 2025) at 2:30pm in Friend 125.
by gradinfo＠cs.princeton.edu 17 Nov '25

17 Nov '25

Abhishek Joshiwill present his MSE talk "Procedural Generation of Articulated Simulation-Ready Assets" TODAY (Monday, November 17, 2025) at 2:30pm in Friend 125. Thesis adviser: Jia Deng, Tom Silver (reader) Abstract: While simulated environments have proven effective for training robots, the lack of high-quality articulated assets makes it difficult for robots to learn new skills involving such objects. To this end, we introduce Infinigen-Articulated, a toolkit for creating realistic, diverse, procedurally generated articulated assets for robot simulation. We introduce procedural generators for 18 common household articulated object categories, along with tools to create custom articulated assets in Blender and export them directly to standard simulation formats including information such as joint dynamics, material physics, collision meshes, and semantic metadata. Experiments demonstrate that assets sampled from these generators are effective for movable part segmentation, training generalizable reinforcement learning policies, and sim-to-real transfer of imitation learning policies.

1 0

CS Colloquium Speaker: Christian Borgs - Monday, November 17 at 12:10pm
by Emily C. Lawrence 17 Nov '25

17 Nov '25

CS Colloquium Speaker Speaker: Christian Borgs, University of California, Berkeley Date: Monday, November 17 Time: 12:10pm EST Location: CS 105 Host: Adji Bousso Dieng Event page: https://www.cs.princeton.edu/events/are-travel-bans-effective-containing-sp… Register for live-stream online here: https://princeton.zoom.us/webinar/register/WN_CGmNJkqmS76ML-byDkxNLA Title: Are Travel Bans Effective in Containing the Spread of a Disease? Abstract: In this talk, I present a mathematical model for the spread of an epidemic from one community to another via travel. Here each community is modeled by a random network (for simplicity, we assume it is an Erdos-Renyi random graph), with the epidemic spread inside the community given by the SIR model on this graph. Travel is modeled by individuals moving from one community to the other at some rate \eta_T, and returning home at another rate \eta_H. We assume that the return rate is of the same order as the recovery rate of the epidemic, while \eta_T is much smaller. Under this assumption, we rigorously prove that if an epidemic starts in the first community, and the second community enacts a travel ban at the moment the epidemic is large enough to be detectable, such a travel ban is ineffective in preventing a large outbreak in the second community. But contrast, other mitigation measure like social distancing or vaccinations (modeled by reducing the rate of infections in the second community) are effective. This is joint work with my students Karissa Huang and Geng Zhao. Bio: Christian Borgs is professor in the Berkeley AI Research Group (BAIR) in the EECS department at UC Berkeley, and faculty director of the Bakar Institute of Digital Materials for the Planet. Borgs was trained as a physicist, holding a Ph.D. from the University of Munich and a Habilitation from the Free University in Berlin, and then becoming chair of Statistical Mechanics at the University of Leipzig. After that, he worked at Microsoft Research for about 20 years, first as co-founder and co-director of the Theory Group, and then as co-founder and Deputy Managing Director of Microsoft Research New England in Cambridge, MA, which was one of the first labs integrating research from areas as diverse as theoretical CS, ML, economics and qualitative social science. He joined Berkeley in 2020. Borgs is a Fellow of the American Mathematical Society, and the American Association for the Advancement of Science. Among the earlier awards he has received are the Karl-Scheel Prize of the German Physical Society, and the Heisenberg Fellowship of the German Research Council. Borgs’ current research focuses on both AI for science and the science of networks, including mathematical foundations, particularly the theory of graph limits aka Graphons (which he co-invented about 15 years ago), graph processes, graph algorithms, and applications of graph theory from economics to systems biology and epidemics. In addition, he has worked on various aspects of responsible AI, from differential privacy to questions of bias in automatic decision making or polarization in online communities. In AI for Science, he is interested in applications of AI to material science, with the aim to create materials to address urgent problems of society like climate change or scarcity of water in many regions of the world.

1 2

Mengzhou Xia will present her FPO "Advancing the Pareto Frontier of Training Open Language Models" on Wednesday, November 19, 2025 at 4:30 PM in CS 105.
by CS Grad Department 14 Nov '25

14 Nov '25

Mengzhou Xia will present her FPO "Advancing the Pareto Frontier of Training Open Language Models" on Wednesday, November 19, 2025 at 4:30 PM in CS 105. The members of Mengzhou’s committee are as follows: Examiners: Danqi Chen (Adviser), Sanjeev Arora, Peter Henderson Readers: Karthik Narasimhan, Pang Wei Koh (University of Washington) A copy of his 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 his talk. Abstract follows below: Large language models (LLMs) have reshaped AI by enabling breakthroughs in language understanding, reasoning, and diverse applications. However, their massive computational demands and the proprietary nature of leading models hinder broad accessibility and customization. My work addresses these challenges by optimizing the use of existing compute, data, and models to push the Pareto frontier for LLM training. In doing so, it not only produces stronger language models but also offers universal approaches that support effective customization and advance our scientific understanding of model training. First, we study how structured pruning can be leveraged to pre-train compact, high performing models at a fraction of the usual pre-training cost, demonstrating its effectiveness in pushing the Pareto frontier for general-purpose pre-training. Next, we turn to the post-training phase to explore the critical role of data in shaping model behavior, presenting principled data optimization techniques that enhance models’ capabilities, safety, and transparency—showing that “less is more” when it comes to constructing effective training datasets. Finally, we introduce novel post-training approaches that more effectively align language models with desired behaviors and objectives. By revealing gaps in the reasoning abilities of even proprietary models, we outline future directions for building AI systems with enhanced reasoning capabilities—focusing on broader data synthesis through agentic processes and enabling advanced applications.

1 0

PPH lunch seminar Nov 14, Mike Skinnider, Accelerating Small Molecule Discovery with Chemical Language Models
by Emily C. Lawrence 14 Nov '25

14 Nov '25

Please join the Princeton Precision Health (PPH) Initiative for a talk for our fall semester series on Friday, November 14th at 12:00 pm, at 252 Nassau Street . Michael Skinnider is an Assistant Professor of the Lewis-Sigler Institute and the Ludwig Princeton Branch and will be presenting a seminar on, Accelerating small molecule discovery with chemical language models. Mass spectrometry-based metabolomics routinely detects thousands of unidentified small molecules within human tissues and biofluids, but structure elucidation of novel metabolites remains a low-throughput endeavour. I will present an approach that leverages chemical language models to discover previously uncharacterized small molecules in biological samples, with applications to the discovery of illicit drugs of abuse and mammalian metabolites. Lunch will be provided. Getting to the seminar space currently requires that you climb a set of stairs. If an accommodation is needed, please contact PPH in advance at: [ mailto:PrincetonPPH@princeton.edu | PrincetonPPH(a)princeton.edu ] Thank you, Princeton Precision Health

1 1

Location change for next week's S.S. Wilks Memorial Seminar in Statistics: Monday, November 17, 2025 at 12:15 PM: Emmanuel Jean Candes, Stanford University
by Emily C. Lawrence 13 Nov '25

13 Nov '25

Next Week’s S.S. Wilks Memorial Seminar in Statistics Date & Time: Monday, November 17, 2025 at 12:15 PM Location: 101 – Friend Center Speaker: [ https://orfe.princeton.edu/events/2025/emmanuel-jean-candes-stanford-univer… | Emmanuel Jean Candes, Stanford University ] Title: What Statistics and AI O ﬀ er Each Other? Abstract: This talk will discuss how thinking carefully about AI inputs and outputs yields more powerful, safer AI. By examining several vignettes, we shall answer questions such as: How do we train language models under cost constraints? What happens when we’ve exhausted all available data? If I start a clinical trial using the drug AI thinks is best, will it pan out? How can we ensure high quality products when AI is used in a larger workflow? That is, how do I know whether AI automated a task correctly? AI powered predictions are beginning to substitute for real data when collection of the latter is difficult, slow, or costly. How then should we leverage machine learning predictions both as a substitute for high-quality data and as a tool for guiding real data collection?

1 0

Next Week's S.S. Wilks Memorial Seminar in Statistics: Monday, November 17, 2025 at 12:15 PM: Emmanuel Jean Candes, Stanford University
by Emily C. Lawrence 13 Nov '25

13 Nov '25

Next Week’s S.S. Wilks Memorial Seminar in Statistics Date & Time: Monday, November 17, 2025 at 12:15 PM Location: 101 - Sherrerd Hall Speaker: [ https://orfe.princeton.edu/events/2025/emmanuel-jean-candes-stanford-univer… | Emmanuel Jean Candes, Stanford University ] Title: What Statistics and AI O ﬀ er Each Other? Abstract: This talk will discuss how thinking carefully about AI inputs and outputs yields more powerful, safer AI. By examining several vignettes, we shall answer questions such as: How do we train language models under cost constraints? What happens when we’ve exhausted all available data? If I start a clinical trial using the drug AI thinks is best, will it pan out? How can we ensure high quality products when AI is used in a larger workflow? That is, how do I know whether AI automated a task correctly? AI powered predictions are beginning to substitute for real data when collection of the latter is difficult, slow, or costly. How then should we leverage machine learning predictions both as a substitute for high-quality data and as a tool for guiding real data collection?

1 0

TODAY at 2:00pm, Special Quantum Seminar: Eli Goldin (NYU), "One-Way Puzzles: A Central Primitive for Quantum Cryptography" (Comp Sci, 105)
by Emily C. Lawrence 10 Nov '25

10 Nov '25

Speaker: Eli Goldin, NYU Quantum Seminar: 2 pm, Mon. Nov. 10, Computer Science Building, CS105 Host: Prof. Alex Lombardi, CS Title: One-Way Puzzles: A Central Primitive for Quantum Cryptography" Abstract: One-way functions are generally seen as a fundamental (or "central") primitive for classical cryptography. The existence of nearly any cryptography implies the existence of one-way functions, and one-way functions satisfy a number of useful properties which make them easy to work with. In the quantum setting, we give evidence that a recently introduced primitive, the one-way puzzle, is a natural analogue of one-way functions in the quantum regime. In particular, they are a useful tool for understanding the fundamental hardness behind quantum cryptography. Towards this end, we show that one-way puzzles satisfy nearly all the "nice properties" of one-way functions. We present a one-way puzzle combiner, a universal construction, a security amplification theorem, and show that one-way puzzles are equivalent to many natural variants. As a result of these nice properties, we demonstrate that one-way puzzles can be constructed from a large class of quantum cryptographic primitives (in particular, those broken by a #P oracle). Furthermore, we show a number of interesting connections between one-way puzzles and the fields of meta-complexity and (efficiently verifiable) quantum advantage, and discuss black-box separations between one-way puzzles and other primitives.

1 0

CS Colloquium Speaker: Ben Shneiderman - Monday, November 10 at 12:10pm
by Emily C. Lawrence 10 Nov '25

10 Nov '25

CS Colloquium Speaker Speaker: Ben Shneiderman, University of Maryland Date: Monday, November 10 Time: 12:10pm EST Location: CS 105 Host: Parastoo Abtahi Event page: https://www.cs.princeton.edu/events/human-centered-ai-augment-amplify-empow… Register for live-stream online here: https://princeton.zoom.us/webinar/register/WN_UIWj3JupSISHG9CTE7M3fw Title: Human-Centered AI: Augment, Amplify, Empower, and Enhance Human Performance Abstract: A new synthesis is emerging that integrates AI technologies with Human-Computer Interaction to produce Human-Centered AI (HCAI). Advocates of this new synthesis seek to amplify, augment, and enhance human abilities, so as to empower people, build their self-efficacy, support creativity, recognize responsibility, and promote social connections. Researchers, developers, business leaders, policy makers and others are expanding the technology-centered scope of Artificial Intelligence (AI) to include Human-Centered AI (HCAI) ways of thinking. This expansion from an algorithm-focused view, which embraces a human-centered perspective, guides evaluation by red team testing, expert reviews, and continuous improvement strategies. These strategies are especially important for Generative AI, which is impressively powerful, but alarmingly flawed. User interface design and governance strategies will be presented. Bio: Ben Shneiderman (http://www.cs.umd.edu/~ben) is an Emeritus Distinguished University Professor in the Department of Computer Science, Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://hcil.umd.edu), and a Member of the UM Institute for Advanced Computer Studies (UMIACS) at the University of Maryland. He is a Fellow of the AAAS, ACM, IEEE, NAI, and the Visualization Academy and a Member of the U.S. National Academy of Engineering. He has received six honorary doctorates in recognition of his pioneering contributions to human-computer interaction and information visualization. His widely-used contributions include the clickable highlighted web-links, high-precision touchscreen keyboards for mobile devices, and tagging for photos. Shneiderman’s information visualization innovations include dynamic query sliders for Spotfire, development of treemaps for viewing hierarchical data, novel network visualizations for NodeXL, and event sequence analysis for electronic health records. Ben is the lead author of Designing the User Interface: Strategies for Effective Human-Computer Interaction (6th ed., 2016). He co-authored Readings in Information Visualization: Using Vision to Think (1999) and Analyzing Social Media Networks with NodeXL (2nd edition, 2019). His book Leonardo’s Laptop (MIT Press) won the IEEE book award for Distinguished Literary Contribution. The New ABCs of Research: Achieving Breakthrough Collaborations (Oxford, 2016) describes how research can produce higher impacts. His book, Human-Centered AI, which was published by Oxford University Press in February 2022, was the winner of the Association of American Publishers award for Computer and Information Systems.

1 1

Ameet Deshpande will present his FPO "From Insights to Improvements: Advancements across the LLM Lifecycle" on Wednesday, November 12, 2025 at 1:30 PM in Louis A. Simpson International Building - A71.
by CS Grad Department 07 Nov '25

07 Nov '25

Ameet Deshpande will present his FPO " From Insights to Improvements: Advancements across the LLM Lifecycle " on Wednesday, November 12, 2025 at 1:30 PM in Louis A. Simpson International Building - A71 . The members of Ameet ’s committee are as follows: Examiners: Karthik Narasimhan (Adviser), Danqi Chen , Sanjeev Arora Readers: Andrés Monroy-Hernández , Suma Bhat A copy of his 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 his talk. Abstract follows below: Large language models (LLMs) have transformed natural language understanding and generation, yet their development remains a fragmented process. Each stage in the LLM lifecycle—pre-training, post-training, evaluation, and improvement—poses distinct challenges that are rarely studied together. This thesis pushes the envelope on all phases while also tying them together for continuous adaptation of models. We begin with the pre-training phase, analyzing what enables cross-lingual generalization in multilingual models, where pre-training on one language bestows skills in another. In the posttraining phase, we show that instruction-tuned LLMs exhibit biases related to the persona they are simulating, highlighting the need for socially aware alignment methods. Turning to evaluation, we demonstrate that existing semantic similarity benchmarks fail to capture contextual reasoning. We propose Conditional STS, a framework that measures similarity relative to explicit conditions, producing fine-grained feedback. Finally, we introduce QUALEVAL, a system that closes the loop between evaluation and improvement by automatically surfacing qualitative insights and guiding targeted model refinement. Together, these contributions reimagine the LLM lifecycle as a continuous feedback process—one where insights from each stage inform the next. By integrating pre-training, alignment, evaluation, and improvement, this thesis moves toward the vision of adaptive, self-improving language models.

1 0