We study the convergence behavior of the celebrated temporal-difference (TD) learning algorithm. By looking at the algorithm through the lens of optimization, we first argue that TD can be viewed as an iterative optimization algorithm where the function to be minimized changes per iteration. By carefully investigating the divergence displayed by TD on a classical counter example, we identify two forces that determine the convergent or divergent behavior of the algorithm. We next formalize our discovery in the linear TD setting with quadratic loss and prove that convergence of TD hinges on the interplay between these two forces. We extend this optimization perspective to prove convergence of TD in a much broader setting than just linear approximation and squared loss. Our results provide a theoretical explanation for the successful application of TD in reinforcement learning.

Deep learning models such as the Transformer are often constructed by heuristics and experience. To provide a complementary foundation, in this work we study the following problem: Is it possible to find an energy function underlying the Transformer model, such that descent steps along this energy correspond with the Transformer forward pass?

Diffusion models have demonstrated empirical successes in various applications and can be adapted to task-specific needs via guidance. This paper studies a form of gradient guidance for adapting a pre-trained diffusion model towards optimizing user-specified objectives. We establish a mathematical framework for guided diffusion to systematically study its optimization theory and algorithmic design. Our theoretical analysis spots a strong link between guided diffusion models and optimization: gradient-guided diffusion models are essentially sampling solutions to a regularized optimization problem, where the regularization is imposed by the pre-training data. As for guidance design, directly bringing in the gradient of an external objective function as guidance would jeopardize the structure in generated samples.

We propose a novel framework for comprehending the reasoning capabilities of large language models (LLMs) through the perspective of meta-learning. By conceptualizing reasoning trajectories as pseudo-gradient descent updates to the LLM's parameters, we identify parallels between LLM reasoning and various meta-learning paradigms. We formalize the training process for reasoning tasks as a meta-learning setup, with each question treated as an individual task, and reasoning trajectories serving as the inner loop optimization for adapting model parameters. Once trained on a diverse set of questions, the LLM develops fundamental reasoning capabilities that can generalize to previously unseen questions. Extensive empirical evaluations substantiate the strong connection between LLM reasoning and meta-learning, exploring several issues of significant interest from a meta-learning standpoint. Our work not only enhances the understanding of LLM reasoning but also provides practical insights for improving these models through established meta-learning techniques.

We study the convergence behavior of the celebrated temporal-difference (TD) learning algorithm. By looking at the algorithm through the lens of optimization, we first argue that TD can be viewed as an iterative optimization algorithm where the function to be minimized changes per iteration. By carefully investigating the divergence displayed by TD on a classical counter example, we identify two forces that determine the convergent or divergent behavior of the algorithm. We next formalize our discovery in the linear TD setting with quadratic loss and prove that convergence of TD hinges on the interplay between these two forces. We extend this optimization perspective to prove convergence of TD in a much broader setting than just linear approximation and squared loss.

Deep learning models such as the Transformer are often constructed by heuristics and experience. To provide a complementary foundation, in this work we study the following problem: Is it possible to find an energy function underlying the Transformer model, such that descent steps along this energy correspond with the Transformer forward pass? This unfolding perspective has been frequently adopted in the past to elucidate more straightforward deep models such as MLPs and CNNs; however, it has thus far remained elusive obtaining a similar equivalence for more complex models with self-attention mechanisms like the Transformer. To this end, we first outline several major obstacles before providing companion techniques to at least partially address them, demonstrating for the first time a close association between energy function minimization and deep layers with self-attention. This interpretation contributes to our intuition and understanding of Transformers, while potentially laying the ground-work for new model designs.

A key challenge in machine learning is to design interpretable models that can reduce their inputs to the best subset for making transparent predictions, especially in the clinical domain. In this work, we propose a certifiably optimal feature selection procedure for logistic regression from a mixed-integer conic optimization perspective that can take an auxiliary cost to obtain features into account. Based on an extensive review of the literature, we carefully create a synthetic dataset generator for clinical prognostic model research. This allows us to systematically evaluate different heuristic and optimal cardinality- and budget-constrained feature selection procedures. The analysis shows key limitations of the methods for the low-data regime and when confronted with label noise. Our paper not only provides empirical recommendations for suitable methods and dataset designs, but also paves the way for future research in the area of meta-learning.

Sequential user behavior modeling plays a crucial role in online user-oriented services, such as product purchasing, news feed consumption, and online advertising. The performance of sequential modeling heavily depends on the scale and quality of historical behaviors. However, the number of user behaviors inherently follows a long-tailed distribution, which has been seldom explored. In this work, we argue that focusing on tail users could bring more benefits and address the long tails issue by learning transferrable parameters from both optimization and feature perspectives. Specifically, we propose a gradient alignment optimizer and adopt an adversarial training scheme to facilitate knowledge transfer from the head to the tail. Such methods can also deal with the cold-start problem of new users. Moreover, it could be directly adaptive to various well-established sequential models. Extensive experiments on four real-world datasets verify the superiority of our framework compared with the state-of-the-art baselines.

Neural architecture search (NAS) -- selecting which neural model to use for your learning problem -- is a promising but computationally expensive direction for automating and democratizing machine learning. The weight-sharing method, whose initial success at dramatically accelerating NAS surprised many in the field, has come under scrutiny due to its poor performance as a surrogate for full model-training (a miscorrelation problem known as rank disorder) and inconsistent results on recent benchmarks. In this post, we give a quick overview of weight-sharing and argue in favor of its continued use for NAS. First-generation NAS methods were astronomically expensive due to the combinatorially large search space, requiring the training of thousands of neural networks to completion. Then, in their 2018 ENAS (for Efficient NAS) paper, Pham et al. introduced the idea of weight-sharing, in which only one shared set of model parameters is trained for all architectures.

Bayesian coresets have emerged as a promising approach for scalable Bayesian inference [22, 12, 13, 11]. The key idea is to select a (weighted) subset of the data such that posterior inference using the selected subset closely approximates posterior inference using the full dataset. This creates a tradeoff, where using Bayesian coresets as opposed to the full dataset exchanges approximation accuracy for computational speedups. We study Bayesian coresets as they are easy to implement, effective in practice, and come with useful theoretical guarantees that relate the coreset size with the approximation quality. The main technical challenge in the Bayesian coreset problem lies in handling the combinatorial constraints - we desire to select a few data points out of many as the coreset. The state of the art approaches rely on two ideas: convexification and greedy methods. In convexification [13], the sparsity constraint - i.e., selection of k data samples - is relaxed into a convex l