Multi-Bin Batching for Increasing LLM Inference Throughput

As large language models (LLMs) grow in popularity for their diverse capabilities, improving the efficiency of their inference systems has become increasingly critical. Batching LLM requests is a critical step in scheduling the inference jobs on servers (e.g. GPUs), enabling the system to maximize throughput by allowing multiple requests to be processed in parallel. However, requests often have varying generation lengths, causing resource underutilization, as hardware must wait for the longest-running request in the batch to complete before moving to the next batch. We formalize this problem from a queueing-theoretic perspective, and aim to design a control policy which is throughput-optimal. We propose Multi-Bin Batching, a simple yet effective method that can provably improve LLM inference throughput by grouping requests with similar (predicted) execution times into predetermined bins. Through a combination of theoretical analysis and experiments, including real-world LLM inference scenarios, we demonstrate significant throughput gains compared to standard batching approaches.

Conflict-Aware Adversarial Training

Adversarial training is the most effective method to obtain adversarial robustness for deep neural networks by directly involving adversarial samples in the training procedure. To obtain an accurate and robust model, the weighted-average method is applied to optimize standard loss and adversarial loss simultaneously. In this paper, we argue that the weighted-average method does not provide the best tradeoff for the standard performance and adversarial robustness. We argue that the failure of the weighted-average method is due to the conflict between the gradients derived from standard and adversarial loss, and further demonstrate such a conflict increases with attack budget theoretically and practically. To alleviate this problem, we propose a new trade-off paradigm for adversarial training with a conflict-aware factor for the convex combination of standard and adversarial loss, named \textbf{Conflict-Aware Adversarial Training~(CA-AT)}. Comprehensive experimental results show that CA-AT consistently offers a superior trade-off between standard performance and adversarial robustness under the settings of adversarial training from scratch and parameter-efficient finetuning.

Communication-Efficient and Tensorized Federated Fine-Tuning of Large Language Models

Parameter-efficient fine-tuning (PEFT) methods typically assume that Large Language Models (LLMs) are trained on data from a single device or client. However, real-world scenarios often require fine-tuning these models on private data distributed across multiple devices. Federated Learning (FL) offers an appealing solution by preserving user privacy, as sensitive data remains on local devices during training. Nonetheless, integrating PEFT methods into FL introduces two main challenges: communication overhead and data heterogeneity. In this paper, we introduce FedTT and FedTT+, methods for adapting LLMs by integrating tensorized adapters into client-side models' encoder/decoder blocks. FedTT is versatile and can be applied to both cross-silo FL and large-scale cross-device FL. FedTT+, an extension of FedTT tailored for cross-silo FL, enhances robustness against data heterogeneity by adaptively freezing portions of tensor factors, further reducing the number of trainable parameters. Experiments on BERT and LLaMA models demonstrate that our proposed methods successfully address data heterogeneity challenges and perform on par or even better than existing federated PEFT approaches while achieving up to 10$\times$ reduction in communication cost.

Long-Term Fairness in Sequential Multi-Agent Selection with Positive Reinforcement

While much of the rapidly growing literature on fair decision-making focuses on metrics for one-shot decisions, recent work has raised the intriguing possibility of designing sequential decision-making to positively impact long-term social fairness. In selection processes such as college admissions or hiring, biasing slightly towards applicants from under-represented groups is hypothesized to provide positive feedback that increases the pool of under-represented applicants in future selection rounds, thus enhancing fairness in the long term. In this paper, we examine this hypothesis and its consequences in a setting in which multiple agents are selecting from a common pool of applicants. We propose the Multi-agent Fair-Greedy policy, that balances greedy score maximization and fairness. Under this policy, we prove that the resource pool and the admissions converge to a long-term fairness target set by the agents when the score distributions across the groups in the population are identical. We provide empirical evidence of existence of equilibria under non-identical score distributions through synthetic and adapted real-world datasets. We then sound a cautionary note for more complex applicant pool evolution models, under which uncoordinated behavior by the agents can cause negative reinforcement, leading to a reduction in the fraction of under-represented applicants. Our results indicate that, while positive reinforcement is a promising mechanism for long-term fairness, policies must be designed carefully to be robust to variations in the evolution model, with a number of open issues that remain to be explored by algorithm designers, social scientists, and policymakers.

Robust Decentralized Learning with Local Updates and Gradient Tracking

As distributed learning applications such as Federated Learning, the Internet of Things (IoT), and Edge Computing grow, it is critical to address the shortcomings of such technologies from a theoretical perspective. As an abstraction, we consider decentralized learning over a network of communicating clients or nodes and tackle two major challenges: data heterogeneity and adversarial robustness. We propose a decentralized minimax optimization method that employs two important modules: local updates and gradient tracking. Minimax optimization is the key tool to enable adversarial training for ensuring robustness. Having local updates is essential in Federated Learning (FL) applications to mitigate the communication bottleneck, and utilizing gradient tracking is essential to proving convergence in the case of data heterogeneity. We analyze the performance of the proposed algorithm, Dec-FedTrack, in the case of nonconvex-strongly concave minimax optimization, and prove that it converges a stationary point. We also conduct numerical experiments to support our theoretical findings.

Generalization Properties of Adversarial Training for $\ell_0$-Bounded Adversarial Attacks

We have widely observed that neural networks are vulnerable to small additive perturbations to the input causing misclassification. In this paper, we focus on the $\ell_0$-bounded adversarial attacks, and aim to theoretically characterize the performance of adversarial training for an important class of truncated classifiers. Such classifiers are shown to have strong performance empirically, as well as theoretically in the Gaussian mixture model, in the $\ell_0$-adversarial setting. The main contribution of this paper is to prove a novel generalization bound for the binary classification setting with $\ell_0$-bounded adversarial perturbation that is distribution-independent. Deriving a generalization bound in this setting has two main challenges: (i) the truncated inner product which is highly non-linear; and (ii) maximization over the $\ell_0$ ball due to adversarial training is non-convex and highly non-smooth. To tackle these challenges, we develop new coding techniques for bounding the combinatorial dimension of the truncated hypothesis class.

Learning to Understand: Identifying Interactions via the Mobius Transform

One of the most fundamental problems in machine learning is finding interpretable representations of the functions we learn. The Mobius transform is a useful tool for this because its coefficients correspond to unique importance scores on sets of input variables. The Mobius Transform is strongly related (and in some cases equivalent) to the concept of Shapley value, which is a widely used game-theoretic notion of importance. This work focuses on the (typical) regime where the fraction of non-zero Mobius coefficients (and thus interactions between inputs) is small compared to the set of all $2^n$ possible interactions between $n$ inputs. When there are $K = O(2^{n \delta})$ with $\delta \leq \frac{1}{3}$ non-zero coefficients chosen uniformly at random, our algorithm exactly recovers the Mobius transform in $O(Kn)$ samples and $O(Kn^2)$ time with vanishing error as $K \rightarrow \infty$, the first non-adaptive algorithm to do so. We also uncover a surprising connection between group testing and the Mobius transform. In the case where all interactions are between at most $t = \Theta(n^{\alpha})$ inputs, for $\alpha < 0.409$, we are able to leverage results from group testing to provide the first algorithm that computes the Mobius transform in $O(Kt\log n)$ sample complexity and $O(K\mathrm{poly}(n))$ time with vanishing error as $K \rightarrow \infty$. Finally, we present a robust version of this algorithm that achieves the same sample and time complexity under some assumptions, but with a factor depending on noise variance. Our work is deeply interdisciplinary, drawing from tools spanning across signal processing, algebra, information theory, learning theory and group testing to address this important problem at the forefront of machine learning.

Inverse Reinforcement Learning by Estimating Expertise of Demonstrators

In Imitation Learning (IL), utilizing suboptimal and heterogeneous demonstrations presents a substantial challenge due to the varied nature of real-world data. However, standard IL algorithms consider these datasets as homogeneous, thereby inheriting the deficiencies of suboptimal demonstrators. Previous approaches to this issue typically rely on impractical assumptions like high-quality data subsets, confidence rankings, or explicit environmental knowledge. This paper introduces IRLEED, Inverse Reinforcement Learning by Estimating Expertise of Demonstrators, a novel framework that overcomes these hurdles without prior knowledge of demonstrator expertise. IRLEED enhances existing Inverse Reinforcement Learning (IRL) algorithms by combining a general model for demonstrator suboptimality to address reward bias and action variance, with a Maximum Entropy IRL framework to efficiently derive the optimal policy from diverse, suboptimal demonstrations. Experiments in both online and offline IL settings, with simulated and human-generated data, demonstrate IRLEED's adaptability and effectiveness, making it a versatile solution for learning from suboptimal demonstrations.

The Fair Value of Data Under Heterogeneous Privacy Constraints

Modern data aggregation often takes the form of a platform collecting data from a network of users. More than ever, these users are now requesting that the data they provide is protected with a guarantee of privacy. This has led to the study of optimal data acquisition frameworks, where the optimality criterion is typically the maximization of utility for the agent trying to acquire the data. This involves determining how to allocate payments to users for the purchase of their data at various privacy levels. The main goal of this paper is to characterize a fair amount to pay users for their data at a given privacy level. We propose an axiomatic definition of fairness, analogous to the celebrated Shapley value. Two concepts for fairness are introduced. The first treats the platform and users as members of a common coalition and provides a complete description of how to divide the utility among the platform and users. In the second concept, fairness is defined only among users, leading to a potential fairness-constrained mechanism design problem for the platform. We consider explicit examples involving private heterogeneous data and show how these notions of fairness can be applied. To the best of our knowledge, these are the first fairness concepts for data that explicitly consider privacy constraints.

Learning-based social coordination to improve safety and robustness of cooperative autonomous vehicles in mixed traffic

It is expected that autonomous vehicles(AVs) and heterogeneous human-driven vehicles(HVs) will coexist on the same road. The safety and reliability of AVs will depend on their social awareness and their ability to engage in complex social interactions in a socially accepted manner. However, AVs are still inefficient in terms of cooperating with HVs and struggle to understand and adapt to human behavior, which is particularly challenging in mixed autonomy. In a road shared by AVs and HVs, the social preferences or individual traits of HVs are unknown to the AVs and different from AVs, which are expected to follow a policy, HVs are particularly difficult to forecast since they do not necessarily follow a stationary policy. To address these challenges, we frame the mixed-autonomy problem as a multi-agent reinforcement learning (MARL) problem and propose an approach that allows AVs to learn the decision-making of HVs implicitly from experience, account for all vehicles' interests, and safely adapt to other traffic situations. In contrast with existing works, we quantify AVs' social preferences and propose a distributed reward structure that introduces altruism into their decision-making process, allowing the altruistic AVs to learn to establish coalitions and influence the behavior of HVs.