Preference-based feedback is important for many applications in reinforcement learning where direct evaluation of a reward function is not feasible. A notable recent example arises in reinforcement learning from human feedback (RLHF) on large language models. For many applications of RLHF, the cost of acquiring the human feedback can be substantial. In this work, we take advantage of the fact that one can often choose contexts at which to obtain human feedback in order to most efficiently identify a good policy, and formalize this as an offline contextual dueling bandit problem. We give an upper-confidence-bound style algorithm for this problem and prove a polynomial worst-case regret bound. We then provide empirical confirmation in a synthetic setting that our approach outperforms existing methods. After, we extend the setting and methodology for practical use in RLHF training of large language models. Here, our method is able to reach better performance with fewer samples of human preferences than multiple baselines on three real-world datasets.
相關內容
Federated learning (FL) systems face performance challenges in dealing with heterogeneous devices and non-identically distributed data across clients. We propose a dynamic global model aggregation method within Asynchronous Federated Learning (AFL) deployments to address these issues. Our aggregation method scores and adjusts the weighting of client model updates based on their upload frequency to accommodate differences in device capabilities. Additionally, we also immediately provide an updated global model to clients after they upload their local models to reduce idle time and improve training efficiency. We evaluate our approach within an AFL deployment consisting of 10 simulated clients with heterogeneous compute constraints and non-IID data. The simulation results, using the FashionMNIST dataset, demonstrate over 10% and 19% improvement in global model accuracy compared to state-of-the-art methods PAPAYA and FedAsync, respectively. Our dynamic aggregation method allows reliable global model training despite limiting client resources and statistical data heterogeneity. This improves robustness and scalability for real-world FL deployments.
Recent advancements in federated learning (FL) have greatly facilitated the development of decentralized collaborative applications, particularly in the domain of Artificial Intelligence of Things (AIoT). However, a critical aspect missing from the current research landscape is the ability to enable data-driven client models with symbolic reasoning capabilities. Specifically, the inherent heterogeneity of participating client devices poses a significant challenge, as each client exhibits unique logic reasoning properties. Failing to consider these device-specific specifications can result in critical properties being missed in the client predictions, leading to suboptimal performance. In this work, we propose a new training paradigm that leverages temporal logic reasoning to address this issue. Our approach involves enhancing the training process by incorporating mechanically generated logic expressions for each FL client. Additionally, we introduce the concept of aggregation clusters and develop a partitioning algorithm to effectively group clients based on the alignment of their temporal reasoning properties. We evaluate the proposed method on two tasks: a real-world traffic volume prediction task consisting of sensory data from fifteen states and a smart city multi-task prediction utilizing synthetic data. The evaluation results exhibit clear improvements, with performance accuracy improved by up to 54% across all sequential prediction models.
Demand forecasting is a prominent business use case that allows retailers to optimize inventory planning, logistics, and core business decisions. One of the key challenges in demand forecasting is accounting for relationships and interactions between articles. Most modern forecasting approaches provide independent article-level predictions that do not consider the impact of related articles. Recent research has attempted addressing this challenge using Graph Neural Networks (GNNs) and showed promising results. This paper builds on previous research on GNNs and makes two contributions. First, we integrate a GNN encoder into a state-of-the-art DeepAR model. The combined model produces probabilistic forecasts, which are crucial for decision-making under uncertainty. Second, we propose to build graphs using article attribute similarity, which avoids reliance on a pre-defined graph structure. Experiments on three real-world datasets show that the proposed approach consistently outperforms non-graph benchmarks. We also show that our approach produces article embeddings that encode article similarity and demand dynamics and are useful for other downstream business tasks beyond forecasting.
We consider training decision trees using noisily labeled data, focusing on loss functions that can lead to robust learning algorithms. Our contributions are threefold. First, we offer novel theoretical insights on the robustness of many existing loss functions in the context of decision tree learning. We show that some of the losses belong to a class of what we call conservative losses, and the conservative losses lead to an early stopping behavior during training and noise-tolerant predictions during testing. Second, we introduce a framework for constructing robust loss functions, called distribution losses. These losses apply percentile-based penalties based on an assumed margin distribution, and they naturally allow adapting to different noise rates via a robustness parameter. In particular, we introduce a new loss called the negative exponential loss, which leads to an efficient greedy impurity-reduction learning algorithm. Lastly, our experiments on multiple datasets and noise settings validate our theoretical insight and the effectiveness of our adaptive negative exponential loss.
A pervasive phenomenon in machine learning applications is distribution shift, where training and deployment conditions for a machine learning model differ. As distribution shift typically results in a degradation in performance, much attention has been devoted to algorithmic interventions that mitigate these detrimental effects. In this paper, we study the effect of distribution shift in the presence of model misspecification, specifically focusing on $L_{\infty}$-misspecified regression and adversarial covariate shift, where the regression target remains fixed while the covariate distribution changes arbitrarily. We show that empirical risk minimization, or standard least squares regression, can result in undesirable misspecification amplification where the error due to misspecification is amplified by the density ratio between the training and testing distributions. As our main result, we develop a new algorithm -- inspired by robust optimization techniques -- that avoids this undesirable behavior, resulting in no misspecification amplification while still obtaining optimal statistical rates. As applications, we use this regression procedure to obtain new guarantees in offline and online reinforcement learning with misspecification and establish new separations between previously studied structural conditions and notions of coverage.
Recent advances in deep learning for physics have focused on discovering shared representations of target systems by incorporating physics priors or inductive biases into neural networks. While effective, these methods are limited to the system domain, where the type of system remains consistent and thus cannot ensure the adaptation to new, or unseen physical systems governed by different laws. For instance, a neural network trained on a mass-spring system cannot guarantee accurate predictions for the behavior of a two-body system or any other system with different physical laws. In this work, we take a significant leap forward by targeting cross domain generalization within the field of Hamiltonian dynamics. We model our system with a graph neural network and employ a meta learning algorithm to enable the model to gain experience over a distribution of tasks and make it adapt to new physics. Our approach aims to learn a unified Hamiltonian representation that is generalizable across multiple system domains, thereby overcoming the limitations of system-specific models. Our results demonstrate that the meta-trained model not only adapts effectively to new systems but also captures a generalized Hamiltonian representation that is consistent across different physical domains. Overall, through the use of meta learning, we offer a framework that achieves cross domain generalization, providing a step towards a unified model for understanding a wide array of dynamical systems via deep learning.
Deep learning could be prone to learning shortcuts raised by dataset bias and result in inaccurate, unreliable, and unfair models, which impedes its adoption in real-world clinical applications. Despite its significance, there is a dearth of research in the medical image classification domain to address dataset bias. Furthermore, the bias labels are often agnostic, as identifying biases can be laborious and depend on post-hoc interpretation. This paper proposes learning Adaptive Agreement from a Biased Council (Ada-ABC), a debiasing framework that does not rely on explicit bias labels to tackle dataset bias in medical images. Ada-ABC develops a biased council consisting of multiple classifiers optimized with generalized cross entropy loss to learn the dataset bias. A debiasing model is then simultaneously trained under the guidance of the biased council. Specifically, the debiasing model is required to learn adaptive agreement with the biased council by agreeing on the correctly predicted samples and disagreeing on the wrongly predicted samples by the biased council. In this way, the debiasing model could learn the target attribute on the samples without spurious correlations while also avoiding ignoring the rich information in samples with spurious correlations. We theoretically demonstrated that the debiasing model could learn the target features when the biased model successfully captures dataset bias. Moreover, to our best knowledge, we constructed the first medical debiasing benchmark from four datasets containing seven different bias scenarios. Our extensive experiments practically showed that our proposed Ada-ABC outperformed competitive approaches, verifying its effectiveness in mitigating dataset bias for medical image classification. The codes and organized benchmark datasets will be made publicly available.
Graph contrastive learning is a general learning paradigm excelling at capturing invariant information from diverse perturbations in graphs. Recent works focus on exploring the structural rationale from graphs, thereby increasing the discriminability of the invariant information. However, such methods may incur in the mis-learning of graph models towards the interpretability of graphs, and thus the learned noisy and task-agnostic information interferes with the prediction of graphs. To this end, with the purpose of exploring the intrinsic rationale of graphs, we accordingly propose to capture the dimensional rationale from graphs, which has not received sufficient attention in the literature. The conducted exploratory experiments attest to the feasibility of the aforementioned roadmap. To elucidate the innate mechanism behind the performance improvement arising from the dimensional rationale, we rethink the dimensional rationale in graph contrastive learning from a causal perspective and further formalize the causality among the variables in the pre-training stage to build the corresponding structural causal model. On the basis of the understanding of the structural causal model, we propose the dimensional rationale-aware graph contrastive learning approach, which introduces a learnable dimensional rationale acquiring network and a redundancy reduction constraint. The learnable dimensional rationale acquiring network is updated by leveraging a bi-level meta-learning technique, and the redundancy reduction constraint disentangles the redundant features through a decorrelation process during learning. Empirically, compared with state-of-the-art methods, our method can yield significant performance boosts on various benchmarks with respect to discriminability and transferability. The code implementation of our method is available at //github.com/ByronJi/DRGCL.
We propose a new method for event extraction (EE) task based on an imitation learning framework, specifically, inverse reinforcement learning (IRL) via generative adversarial network (GAN). The GAN estimates proper rewards according to the difference between the actions committed by the expert (or ground truth) and the agent among complicated states in the environment. EE task benefits from these dynamic rewards because instances and labels yield to various extents of difficulty and the gains are expected to be diverse -- e.g., an ambiguous but correctly detected trigger or argument should receive high gains -- while the traditional RL models usually neglect such differences and pay equal attention on all instances. Moreover, our experiments also demonstrate that the proposed framework outperforms state-of-the-art methods, without explicit feature engineering.
Deep neural networks (DNNs) have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs. Such adversarial examples can mislead DNNs to produce adversary-selected results. Different attack strategies have been proposed to generate adversarial examples, but how to produce them with high perceptual quality and more efficiently requires more research efforts. In this paper, we propose AdvGAN to generate adversarial examples with generative adversarial networks (GANs), which can learn and approximate the distribution of original instances. For AdvGAN, once the generator is trained, it can generate adversarial perturbations efficiently for any instance, so as to potentially accelerate adversarial training as defenses. We apply AdvGAN in both semi-whitebox and black-box attack settings. In semi-whitebox attacks, there is no need to access the original target model after the generator is trained, in contrast to traditional white-box attacks. In black-box attacks, we dynamically train a distilled model for the black-box model and optimize the generator accordingly. Adversarial examples generated by AdvGAN on different target models have high attack success rate under state-of-the-art defenses compared to other attacks. Our attack has placed the first with 92.76% accuracy on a public MNIST black-box attack challenge.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191