39 citations · 175 across the 48 of their papers we have counts for
30 papers · 1 filter
Score-Calibrated Flow for Sampling from Unnormalized Densities with Applications to Generative Online Reinforcement Learning
Zeyang Li, Yunan Wang, Risheek Garrepalli +2
Diffusion and flow models provide expressive policy classes for online reinforcement learning (RL), enabling multimodal behaviors and improved performance. However, training these…
Safe Meta-Reinforcement Learning via Information Space Reachability
Zeyang Li, Sunbochen Tang, Navid Azizan
Meta-reinforcement learning (meta-RL) enables agents to adapt to unseen tasks with limited experience. Despite its promise, the application of meta-RL in real-world tasks is hinder…
Newton Matching for Generative Modeling: A Unified Framework for Fine-Tuning and Sampling
Zeyang Li, Yunan Wang, Paolo Giaretta +1
We develop Newton Matching, a unified framework for fine-tuning and sampling in generative modeling. The target is , where is the reward, the inverse tem…
DiRecT: Safe Diffusion-Based Planning via Receding-Horizon Denoising
Paolo Giaretta, Zeyang Li, Navid Azizan
Diffusion models have emerged as powerful tools for planning and control by learning multimodal distributions over actions and trajectories. Yet reliable inference-time safety enfo…
Provably Safe, Yet Scalable Reinforcement Learning
Kai S. Yun, Zeyang Li, Navid Azizan
Safe reinforcement learning (RL) aims to learn policies that optimize rewards while satisfying constraints. Predominant approaches rely on soft-constrained policy optimization, whi…
HardNet++: Nonlinear Constraint Enforcement in Neural Networks
Andrea Goertzen, Kaveh Alim, Youngjae Min +1
Enforcing constraint satisfaction in neural network outputs is critical for safety, reliability, and physical fidelity in many control and decision-making applications. While soft-…
LMI-Net: Linear Matrix Inequality--Constrained Neural Networks via Differentiable Projection Layers
Sunbochen Tang, Andrea Goertzen, Navid Azizan
Linear matrix inequalities (LMIs) have played a central role in certifying stability, robustness, and forward invariance of dynamical systems. Despite rapid development in learning…
Reverse Flow Matching: A Unified Framework for Online Reinforcement Learning with Diffusion and Flow Policies
Zeyang Li, Sunbochen Tang, Navid Azizan
Diffusion and flow policies are gaining prominence in online reinforcement learning (RL) due to their expressive power, yet training them efficiently remains a critical challenge.…
HardFlow: Hard-Constrained Sampling for Flow-Matching Models via Trajectory Optimization
Zeyang Li, Kaveh Alim, Navid Azizan
Diffusion and flow-matching have emerged as powerful methodologies for generative modeling, with remarkable success in capturing complex data distributions and enabling flexible gu…
Probabilistic Forecasting for Building Energy Systems using Time-Series Foundation Models
Young Jin Park, Francois Germain, Jing Liu +6
Decision-making in building energy systems critically depends on the predictive accuracy of relevant time-series models. In scenarios lacking extensive data from a target building,…
On the Role of Transformer Feed-Forward Layers in Nonlinear In-Context Learning
Haoyuan Sun, Ali Jadbabaie, Navid Azizan
Transformer-based models demonstrate a remarkable ability for in-context learning (ICL), where they can adapt to unseen tasks from a few prompt examples without parameter updates.…
Safe Multi-Agent Reinforcement Learning with Convergence to Generalized Nash Equilibrium
Zeyang Li, Navid Azizan
Multi-agent reinforcement learning (MARL) has achieved notable success in cooperative tasks, demonstrating impressive performance and scalability. However, deploying MARL agents in…
Optimizing Attention with Mirror Descent: Generalized Max-Margin Token Selection
Addison Kristanto Julistiono, Davoud Ataee Tarzanagh, Navid Azizan
Attention mechanisms have revolutionized several domains of artificial intelligence, such as natural language processing and computer vision, by enabling models to selectively focu…
HardNet: Hard-Constrained Neural Networks with Universal Approximation Guarantees
Youngjae Min, Navid Azizan
Incorporating prior knowledge or specifications of input-output relationships into machine learning models has attracted significant attention, as it enhances generalization from l…
Differentially Private Synthetic Data Generation for Relational Databases
Kaveh Alimohammadi, Hao Wang, Ojas Gulati +2
Existing differentially private (DP) synthetic data generation mechanisms typically assume a single-source table. In practice, data is often distributed across multiple tables with…
A Unified Approach to Controlling Implicit Regularization via Mirror Descent
Haoyuan Sun, Khashayar Gatmiry, Kwangjun Ahn +1
Inspired by the remarkable success of large neural networks, there has been significant interest in understanding the generalization performance of over-parameterized models. Subst…
Training-Free Uncertainty Estimation for Embedding Models
Young-Jin Park, Hao Wang, Addison Kristanto Julistiono +2
Embedding models, often obtained via self-supervised learning, extract general-purpose representations from data. Quantifying the reliability of these representations is crucial, a…
SketchOGD: Memory-Efficient Continual Learning
Youngjae Min, Benjamin Wright, Jeremy Bernstein +1
When machine learning models are trained continually on a sequence of tasks, they are often liable to forget what they learned on previous tasks--a phenomenon known as catastrophic…
Automatic Gradient Descent: Deep Learning without Hyperparameters
Jeremy Bernstein, Chris Mingard, Kevin Huang +2
The architecture of a deep neural network is defined explicitly in terms of the number of layers, the width of each layer and the general network topology. Existing optimisation fr…
Tractable Uncertainty-Aware Meta-Learning
Young-Jin Park, Cesar Almecija, Apoorva Sharma +1
Meta-learning is a popular approach for learning new tasks with limited data by leveraging the commonalities among different tasks. However, meta-learned models can perform poorly…
Uncertainty in Contrastive Learning: On the Predictability of Downstream Performance
Shervin Ardeshir, Navid Azizan
The superior performance of some of today's state-of-the-art deep learning models is to some extent owed to extensive (self-)supervised contrastive pretraining on large-scale datas…
ORFit: One-Pass Learning via Bridging Orthogonal Gradient Descent and Recursive Least-Squares
Youngjae Min, Namhoon Cho, Navid Azizan
While large machine learning models have shown remarkable performance in various domains, their training typically requires iterating for many passes over the training data. Howeve…
Mirror Descent Maximizes Generalized Margin and Can Be Implemented Efficiently
Haoyuan Sun, Kwangjun Ahn, Christos Thrampoulidis +1
Driven by the empirical success and wide use of deep neural networks, understanding the generalization performance of overparameterized models has become an increasingly popular qu…
Online Learning for Traffic Routing under Unknown Preferences
Devansh Jalota, Karthik Gopalakrishnan, Navid Azizan +2
In transportation networks, users typically choose routes in a decentralized and self-interested manner to minimize their individual travel costs, which, in practice, often results…