Publications (29)
DischargeSim: A Simulation Benchmark for Educational Doctor-Patient Communication at Discharge
Zonghai Yao, Michael Sun, Won Seok Jang +3
Discharge communication is a critical yet underexplored component of patient care, where the goal shifts from diagnosis to education. While recent large language model (LLM) benchm…
Study of Exoo's Lower Bound for Ramsey number
Lachlan Ge, Yasiru Jayasooriya, Alex Qiu +2
We review Exoo's 1989 paper, which demonstrates that a lower bound for the Ramsey number is . We provide an efficient way to verify the claims in the paper, adding det…
Eigenvalue-less matrices from SL to PSL
Juan Lucas Callo, George Chen, Yasiru Jayasooriya +5
We analyse the set of matrices in SL without eigenvalues explicitly, extracting nice bijections between the 18 equally sized conjugacy classes contained within. I…
Multimodal Large Language Models for Inverse Molecular Design with Retrosynthetic Planning
Gang Liu, Michael Sun, Wojciech Matusik +2
While large language models (LLMs) have integrated images, adapting them to graphs remains challenging, limiting their applications in materials and drug design. This difficulty st…
Representing Molecules as Random Walks Over Interpretable Grammars
Michael Sun, Minghao Guo, Weize Yuan +10
Recent research in molecular discovery has primarily been devoted to small, drug-like molecules, leaving many similarly important applications in material design without adequate t…
Almost-Nash Sequential Bargaining
Gokul Dharan, Hunter Guru, Michael Sun
In a 2017 paper, later presented at the Web and Internet Economics conference, titled ``Sequential Deliberation for Social Choice", the authors propose a mechanism in which a serie…
Protein Structure Tokenization via Geometric Byte Pair Encoding
Michael Sun, Weize Yuan, Gang Liu +2
Protein structure is central to biological function, and enabling multimodal protein models requires joint reasoning over sequence, structure, and function. A key barrier is the la…
Translation of Mathieu's On the five-fold transitive function of 24 quantities
Yiming Bing, Bright Hu, Ronni Hu +11
We translated the paper 'On the five-fold transitive function of 24 quantities' by Ãmile Mathieu into English. This was done to aid our project on expressing as additive…
Understanding Fermat's Last Theorem's Proofs
Alex Qiu, Tanish Sarathy, Spencer Nicklin +1
We take the perspective of an advanced high school student trying to understand the proof of Fermat's Last Theorem for the first time. We collect definitions and statements needed…
Directed Graph Grammars for Sequence-based Learning
Michael Sun, Orion Foo, Gang Liu +2
Directed acyclic graphs (DAGs) are a class of graphs commonly used in practice, with examples that include electronic circuits, Bayesian networks, and neural architectures. While m…
Autonomous Agents for Scientific Discovery: Orchestrating Scientists, Language, Code, and Physics
Lianhao Zhou, Hongyi Ling, Cong Fu +14
Computing has long served as a cornerstone of scientific discovery. Recently, a paradigm shift has emerged with the rise of large language models (LLMs), introducing autonomous sys…
Preventing Catastrophic Forgetting: Behavior-Aware Sampling for Safer Language Model Fine-Tuning
Anh Pham, Mihir Thalanki, Michael Sun +6
Large language models often lose previously aligned safety behaviors when fine-tuned on benign data, a phenomenon known as catastrophic forgetting. Prior work shows that adding ran…
Graph Diffusion Transformers are In-Context Molecular Designers
Gang Liu, Jie Chen, Yihan Zhu +4
In-context learning allows large models to adapt to new tasks from a few demonstrations, but it has shown limited success in molecular design. Existing databases such as ChEMBL con…
The Mathieu group as additive functions on the finite field of size
Yiming Bing, Bright Hu, Ronni Hu +8
We explicitly extend the standard permutation action of the Mathieu group on a 23 element set contained in a finite field of elements $\mathbb{F}_{2^{1…
Foundation Molecular Grammar: Multi-Modal Foundation Models Induce Interpretable Molecular Graph Languages
Michael Sun, Weize Yuan, Gang Liu +2
Recent data-efficient molecular generation approaches exploit graph grammars to introduce interpretability into the generative models. However, grammar learning therein relies on e…
Two-Stage Pretraining for Molecular Property Prediction in the Wild
Kevin Tirta Wijaya, Minghao Guo, Michael Sun +3
Molecular deep learning models have achieved remarkable success in property prediction, but they often require large amounts of labeled data. The challenge is that, in real-world a…
Procedural Synthesis of Synthesizable Molecules
Michael Sun, Alston Lo, Minghao Guo +3
Designing synthetically accessible molecules and recommending analogs to unsynthesizable molecules are important problems for accelerating molecular discovery. We reconceptualize b…
Level-rank duality via tensor categories
Victor Ostrik, Michael Sun
We give a new way to derive branching rules for the conformal embedding $$(\asl_n)_m\oplus(\asl_m)_n\subset(\asl_{nm})_1.$$ In addition, we show that the category $\Cc(\asl_n)_m^0$…
Bounds on metric dimension for families of planar graphs
Carl Joshua Quines, Michael Sun
The concept of metric dimension has applications in a variety of fields, such as chemistry, robotic navigation, and combinatorial optimization. We show bounds for graphs with v…
Privacy Preserving Inference of Personalized Content for Out of Matrix Users
Michael Sun, Tai Vu, Andrew Wang
Recommender systems for niche and dynamic communities face persistent challenges from data sparsity, cold start users and items, and privacy constraints. Traditional collaborative…
Post Hoc Regression Refinement via Pairwise Rankings
Kevin Tirta Wijaya, Michael Sun, Minghao Guo +3
Accurate prediction of continuous properties is essential to many scientific and engineering tasks. Although deep-learning regressors excel with abundant labels, their accuracy det…
Improving Representational Continuity via Continued Pretraining
Michael Sun, Ananya Kumar, Divyam Madaan +1
We consider the continual representation learning setting: sequentially pretrain a model on tasks , and then adapt on a small amount of data from each t…
Do Neural Networks Generalize from Self-Averaging Sub-classifiers in the Same Way As Adaptive Boosting?
Michael Sun, Peter Chatain
In recent years, neural networks (NNs) have made giant leaps in a wide variety of domains. NNs are often referred to as black box algorithms due to how little we can explain their…
Investigating Effect of Dialogue History in Multilingual Task Oriented Dialogue Systems
Michael Sun, Kaili Huang, Mehrad Moradshahi
While the English virtual assistants have achieved exciting performance with an enormous amount of training resources, the needs of non-English-speakers have not been satisfied wel…
Security Analysis of Near-Field Communication (NFC) Payments
Dennis Giese, Kevin Liu, Michael Sun +2
Near-Field Communication (NFC) is a modern technology for short range communication with a variety of applications ranging from physical access control to contactless payments. The…
Symplectic level-rank duality via tensor categories
Victor Ostrik, Eric C. Rowell, Michael Sun
We give two proofs of a level-rank duality for braided fusion categories obtained from quantum groups of type at roots of unity. The first proof uses conformal embeddings, whil…
Alpha-Mini: Minichess Agent with Deep Reinforcement Learning
Michael Sun, Robert Tan
We train an agent to compete in the game of Gardner minichess, a downsized variation of chess played on a 5x5 board. We motivated and applied a SOTA actor-critic method Proximal Po…
Model actions for almost reduced groups on UHF algebras
Michael Sun
For any countable discrete group with a reduced abelian subgroup of finite index, we construct an action of on the universal UHF algebra $\Qq$ using an infinite tensor…
X-RiSAWOZ: High-Quality End-to-End Multilingual Dialogue Datasets and Few-shot Agents
Mehrad Moradshahi, Tianhao Shen, Kalika Bali +16
Task-oriented dialogue research has mainly focused on a few popular languages like English and Chinese, due to the high dataset creation cost for a new language. To reduce the cost…