6 citations · 6 across the 15 of their papers we have counts for
17 papers
ORCA: The Annealed Spectral Conditioning Optimizer for Faster, Better LLM Training
Yuanshi Liu, Boyuan Jiang, Liang Hou +4
Modern LLM optimizers such as Muon often produce weight matrices with higher effective rank than Adam, yet further spectral control has delivered only modest gains. We identify a t…
Beyond Overparameterization: Provable Learning of Input-Convex Multi-Layer Polynomial Networks with Active Queries
Jinqi Tang, Qian Chen, Shihong Ding +1
The theoretical understanding of multi-layer neural networks is largely confined to overparameterized settings, which obscure parameter identifiability and incur high sample comple…
What Makes a Strong Model? A Unified Spectral Analysis of Knowledge Transfer over High-dimensional Linear Regression
Wendao Wu, Fangqing Zhang, Haihan Zhang +1
Teacher-Student Knowledge Transfer (KT) is ubiquitous in modern machine learning, ranging from classical model compression via Knowledge Distillation (KD) to the emergent phenomeno…
SPHERE: Mitigating the Loss of Spectral Plasticity in Mixture-of-Experts for Deep Reinforcement Learning
Lirui Luo, Guoxi Zhang, Hongming Xu +2
In deep reinforcement learning (DRL), an agent is trained from a stream of experience. In a continual learning setting, such agents can suffer from plasticity loss: their ability t…
Near-optimal and Efficient First-Order Algorithm for Multi-Task Learning with Shared Linear Representation
Shihong Ding, Fangyu Du, Cong Fang
Multi-task learning (MTL) has emerged as a pivotal paradigm in machine learning by leveraging shared structures across multiple related tasks. Despite its empirical success, the de…
Mild Over-Parameterization Benefits Asymmetric Tensor PCA
Shihong Ding, Weicheng Lin, Cong Fang
Asymmetric Tensor PCA (ATPCA) is a prototypical model for studying the trade-offs between sample complexity, computation, and memory. Existing algorithms for this problem typically…
Accelerating Single-Pass SGD for Generalized Linear Prediction
Qian Chen, Shihong Ding, Cong Fang
We study generalized linear prediction under a streaming setting, where each iteration uses only one fresh data point for a gradient-level update. While momentum is well-establishe…
MVR: Multi-view Video Reward Shaping for Reinforcement Learning
Lirui Luo, Guoxi Zhang, Hongming Xu +3
Reward design is of great importance for solving complex tasks with reinforcement learning. Recent studies have explored using image-text similarity produced by vision-language mod…
Near-Optimal Tensor PCA via Normalized Stochastic Gradient Ascent with Overparameterization
Shihong Ding, Yihong Gu, Yuanshi Liu +1
We study the Order- () spiked tensor model for the tensor principal component analysis (PCA) problem: given i.i.d. observations of a -th order tensor generated…
AdaPM: a Partial Momentum Algorithm for LLM Training
Yimu Zhang, Yuanshi Liu, Cong Fang
In the training of large language models, momentum is widely used and often demonstrated to achieve significant acceleration. However, storing momentum typically presents memory ch…
Learning Curves of Stochastic Gradient Descent in Kernel Regression
Haihan Zhang, Weicheng Lin, Yuanshi Liu +1
This paper considers a canonical problem in kernel regression: how good are the model performances when it is trained by the popular online first-order algorithms, compared to the…
Scaling Law for Stochastic Gradient Descent in Quadratically Parameterized Linear Regression
Shihong Ding, Haihan Zhang, Hanzhen Zhao +1
In machine learning, the scaling law describes how the model performance improves with the model and data size scaling up. From a learning theory perspective, this class of results…
Optimal Algorithms in Linear Regression under Covariate Shift: On the Importance of Precondition
Yuanshi Liu, Haihan Zhang, Qian Chen +1
A common pursuit in modern statistical learning is to attain satisfactory generalization out of the source data distribution (OOD). In theory, the challenge remains unsolved even u…
The Optimality of (Accelerated) SGD for High-Dimensional Quadratic Optimization
Haihan Zhang, Yuanshi Liu, Qianwen Chen +1
Stochastic gradient descent (SGD) is a widely used algorithm in machine learning, particularly for neural network training. Recent studies on SGD for canonical quadratic optimizati…
Optimizing over Multiple Distributions under Generalized Quasar-Convexity Condition
Shihong Ding, Long Yang, Luo Luo +1
We study a typical optimization model where the optimization variable is composed of multiple probability distributions. Though the model appears frequently in practice, such as fo…
Quantum Algorithms and Lower Bounds for Finite-Sum Optimization
Yexin Zhang, Chenyi Zhang, Cong Fang +2
Finite-sum optimization has wide applications in machine learning, covering important problems such as support vector machines, regression, etc. In this paper, we initiate the stud…
On the Algorithmic Bias of Aligning Large Language Models with RLHF: Preference Collapse and Matching Regularization
Jiancong Xiao, Ziniu Li, Xingyu Xie +4
Accurately aligning large language models (LLMs) with human preferences is crucial for informing fair, economically sound, and statistically efficient decision-making processes. Ho…