Developing an agent capable of adapting to unseen environments remains a difficult challenge in imitation learning. This work presents Adaptive Return-conditioned Policy (ARP), an efficient framework designed to enhance the agent's generalization ability using natural language task descriptions and pre-trained multimodal encoders. Our key idea is to calculate a similarity between visual observations and natural language instructions in the pre-trained multimodal embedding space (such as CLIP) and use it as a reward signal. We then train a return-conditioned policy using expert demonstrations labeled with multimodal rewards. Because the multimodal rewards provide adaptive signals at each timestep, our ARP effectively mitigates the goal misgeneralization. This results in superior generalization performances even when faced with unseen text instructions, compared to existing text-conditioned policies. To improve the quality of rewards, we also introduce a fine-tuning method for pre-trained multimodal encoders, further enhancing the performance. Video demonstrations and source code are available on the project website: \url{//sites.google.com/view/2023arp}.
相關內容
We develop a more flexible approach for identifying and estimating average counterfactual outcomes when several but not all possible outcomes are observed for each unit in a large cross section. Such settings include event studies and studies of outcomes of "matches" between agents of two types, e.g. workers and firms or people and places. When outcomes are generated by a factor model that allows for low-dimensional unobserved confounders, our method yields consistent, asymptotically normal estimates of counterfactual outcome means under asymptotics that fix the number of outcomes as the cross section grows and general outcome missingness patterns, including those not accommodated by existing methods. Our method is also computationally efficient, requiring only a single eigendecomposition of a particular aggregation of any factor estimates constructed using subsets of units with the same observed outcomes. In a semi-synthetic simulation study based on matched employer-employee data, our method performs favorably compared to a Two-Way-Fixed-Effects-model-based estimator.
Few-shot learning has been studied to adapt models to tasks with very few samples. It holds profound significance, particularly in clinical tasks, due to the high annotation cost of medical images. Several works have explored few-shot learning on medical images, yet they still require a large number of medical images for pre-training models to gain domain-specific priors. Vision foundation models recently have achieved remarkable success in natural images. Hence, adapting rapidly advancing vision foundation models from natural images to few-shot clinical tasks holds great promise. MedFMC has recently organized a challenge to shed more light on this topic at NeurIPS 2023. In this work, we present our challenge solution. We observe that a simple variant of fine-tuning with partial freezing shows remarkable performance. Empirical evidence demonstrates that this approach could outperform various common fine-tuning methods under limited sample sizes. Additionally, we explore enhanced utilization of semantic supervision to boost performance. We propose a novel approach that contextualizes labels via large language models (LLMs). Our findings reveal that the context generated by LLMs significantly enhances the discrimination of semantic embeddings for similar categories, resulting in a notable performance improvement of 3%-5% in 1-shot settings compared to commonly employed one-hot labels and other semantic supervision methods. Our solution secures the 1st place in the MedFMC challenge.
Deep learning succeeds by doing hierarchical feature learning, yet tuning Hyper-Parameters (HP) such as initialization scales, learning rates etc., only give indirect control over this behavior. In this paper, we propose the alignment between the feature updates and the backward pass as a key notion to predict, measure and control feature learning. On the one hand, we show that when alignment holds, the magnitude of feature updates after one SGD step is related to the magnitude of the forward and backward passes by a simple and general formula. This leads to techniques to automatically adjust HPs (initialization scales and learning rates) at initialization and throughout training to attain a desired feature learning behavior. On the other hand, we show that, at random initialization, this alignment is determined by the spectrum of a certain kernel, and that well-conditioned layer-to-layer Jacobians (aka dynamical isometry) implies alignment. Finally, we investigate ReLU MLPs and ResNets in the large width-then-depth limit. Combining hints from random matrix theory and numerical experiments, we show that (i) in MLP with iid initializations, alignment degenerates with depth, making it impossible to start training, and that (ii) in ResNets, the branch scale $1/\sqrt{\text{depth}}$ is the only one maintaining non-trivial alignment at infinite depth.
We study robust reinforcement learning (RL) with the goal of determining a well-performing policy that is robust against model mismatch between the training simulator and the testing environment. Previous policy-based robust RL algorithms mainly focus on the tabular setting under uncertainty sets that facilitate robust policy evaluation, but are no longer tractable when the number of states scales up. To this end, we propose two novel uncertainty set formulations, one based on double sampling and the other on an integral probability metric. Both make large-scale robust RL tractable even when one only has access to a simulator. We propose a robust natural actor-critic (RNAC) approach that incorporates the new uncertainty sets and employs function approximation. We provide finite-time convergence guarantees for the proposed RNAC algorithm to the optimal robust policy within the function approximation error. Finally, we demonstrate the robust performance of the policy learned by our proposed RNAC approach in multiple MuJoCo environments and a real-world TurtleBot navigation task.
Logical frameworks are successful in modeling proof systems. Recently, CoLF extended the logical framework LF to support higher-order rational terms that enable adequate encoding of circular objects and derivations. In this paper, we propose CoLF$^\omega$ as an alternative interpretation of CoLF-style signatures where terms are taken to be all possibly infinitary terms that are consistent with a given signature. In particular, we propose the notion of productive B\"ohm trees, a particular kind of typed $\bot$-free B\"ohm trees that are closed under hereditary substitution. We show that the productive B\"ohm trees are capable of meta-encoding their own structure. Overall, we hope to establish CoLF$^\omega$ as a new formal framework for the encoding of infinitary regular and non-regular structures.
Code optimization is a daunting task that requires a significant level of expertise from experienced programmers. This level of expertise is not sufficient when compared to the rapid development of new hardware architectures. Towards advancing the whole code optimization process, recent approaches rely on machine learning and artificial intelligence techniques. This paper introduces a new framework to decrease the complexity of code optimization. The proposed framework builds on large language models (LLMs) and reinforcement learning (RL) and enables LLMs to receive feedback from their environment (i.e., unit tests) during the fine-tuning process. We compare our framework with existing state-of-the-art models and show that it is more efficient with respect to speed and computational usage, as a result of the decrement in training steps and its applicability to models with fewer parameters. Additionally, our framework reduces the possibility of logical and syntactical errors. Toward evaluating our approach, we run several experiments on the PIE dataset using a CodeT5 language model and RRHF, a new reinforcement learning algorithm. We adopt a variety of evaluation metrics with regards to optimization quality, and speedup. The evaluation results demonstrate that the proposed framework has similar results in comparison with existing models using shorter training times and smaller pre-trained models. In particular, we accomplish an increase of 5.6% and 2.2 over the baseline models concerning the %OP T and SP metrics.
Sequential recommendation aims to leverage users' historical behaviors to predict their next interaction. Existing works have not yet addressed two main challenges in sequential recommendation. First, user behaviors in their rich historical sequences are often implicit and noisy preference signals, they cannot sufficiently reflect users' actual preferences. In addition, users' dynamic preferences often change rapidly over time, and hence it is difficult to capture user patterns in their historical sequences. In this work, we propose a graph neural network model called SURGE (short for SeqUential Recommendation with Graph neural nEtworks) to address these two issues. Specifically, SURGE integrates different types of preferences in long-term user behaviors into clusters in the graph by re-constructing loose item sequences into tight item-item interest graphs based on metric learning. This helps explicitly distinguish users' core interests, by forming dense clusters in the interest graph. Then, we perform cluster-aware and query-aware graph convolutional propagation and graph pooling on the constructed graph. It dynamically fuses and extracts users' current activated core interests from noisy user behavior sequences. We conduct extensive experiments on both public and proprietary industrial datasets. Experimental results demonstrate significant performance gains of our proposed method compared to state-of-the-art methods. Further studies on sequence length confirm that our method can model long behavioral sequences effectively and efficiently.
Recently, contrastive learning (CL) has emerged as a successful method for unsupervised graph representation learning. Most graph CL methods first perform stochastic augmentation on the input graph to obtain two graph views and maximize the agreement of representations in the two views. Despite the prosperous development of graph CL methods, the design of graph augmentation schemes -- a crucial component in CL -- remains rarely explored. We argue that the data augmentation schemes should preserve intrinsic structures and attributes of graphs, which will force the model to learn representations that are insensitive to perturbation on unimportant nodes and edges. However, most existing methods adopt uniform data augmentation schemes, like uniformly dropping edges and uniformly shuffling features, leading to suboptimal performance. In this paper, we propose a novel graph contrastive representation learning method with adaptive augmentation that incorporates various priors for topological and semantic aspects of the graph. Specifically, on the topology level, we design augmentation schemes based on node centrality measures to highlight important connective structures. On the node attribute level, we corrupt node features by adding more noise to unimportant node features, to enforce the model to recognize underlying semantic information. We perform extensive experiments of node classification on a variety of real-world datasets. Experimental results demonstrate that our proposed method consistently outperforms existing state-of-the-art baselines and even surpasses some supervised counterparts, which validates the effectiveness of the proposed contrastive framework with adaptive augmentation.
It is important to detect anomalous inputs when deploying machine learning systems. The use of larger and more complex inputs in deep learning magnifies the difficulty of distinguishing between anomalous and in-distribution examples. At the same time, diverse image and text data are available in enormous quantities. We propose leveraging these data to improve deep anomaly detection by training anomaly detectors against an auxiliary dataset of outliers, an approach we call Outlier Exposure (OE). This enables anomaly detectors to generalize and detect unseen anomalies. In extensive experiments on natural language processing and small- and large-scale vision tasks, we find that Outlier Exposure significantly improves detection performance. We also observe that cutting-edge generative models trained on CIFAR-10 may assign higher likelihoods to SVHN images than to CIFAR-10 images; we use OE to mitigate this issue. We also analyze the flexibility and robustness of Outlier Exposure, and identify characteristics of the auxiliary dataset that improve performance.
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.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191