Investigating Pre-Training Objectives for Generalization in Vision-Based Reinforcement Learning

• URL: http://arxiv.org/abs/2406.06037v1
• Date: Mon, 10 Jun 2024 06:06:38 GMT
• Title: Investigating Pre-Training Objectives for Generalization in Vision-Based Reinforcement Learning
• Authors: Donghu Kim, Hojoon Lee, Kyungmin Lee, Dongyoon Hwang, Jaegul Choo,
• Abstract summary: We introduce the Atari Pre-training Benchmark (Atari-PB), which pre-trains a ResNet-50 model on 10 million transitions from 50 Atari games. Our experiments show that pre-training objectives focused on learning task-agnostic features enhance generalization across different environments. In contrast, objectives focused on learning task-specific knowledge improve performance in environments similar to the pre-training dataset but not in varied ones.
• Score: 41.1883663165909
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recently, various pre-training methods have been introduced in vision-based Reinforcement Learning (RL). However, their generalization ability remains unclear due to evaluations being limited to in-distribution environments and non-unified experimental setups. To address this, we introduce the Atari Pre-training Benchmark (Atari-PB), which pre-trains a ResNet-50 model on 10 million transitions from 50 Atari games and evaluates it across diverse environment distributions. Our experiments show that pre-training objectives focused on learning task-agnostic features (e.g., identifying objects and understanding temporal dynamics) enhance generalization across different environments. In contrast, objectives focused on learning task-specific knowledge (e.g., identifying agents and fitting reward functions) improve performance in environments similar to the pre-training dataset but not in varied ones. We publicize our codes, datasets, and model checkpoints at https://github.com/dojeon-ai/Atari-PB.

Related papers

• Zero-Shot Object-Centric Representation Learning [72.43369950684057]
  We study current object-centric methods through the lens of zero-shot generalization. We introduce a benchmark comprising eight different synthetic and real-world datasets. We find that training on diverse real-world images improves transferability to unseen scenarios.
  arXiv  Detail & Related papers  (2024-08-17T10:37:07Z)
• DRED: Zero-Shot Transfer in Reinforcement Learning via Data-Regularised Environment Design [11.922951794283168]
  In this work, we investigate how the sampling of individual environment instances, or levels, affects the zero-shot generalisation (ZSG) ability of RL agents. We discover that for deep actor-critic architectures sharing their base layers, prioritising levels according to their value loss minimises the mutual information between the agent's internal representation and the set of training levels in the generated training data. We find that existing UED methods can significantly shift the training distribution, which translates to low ZSG performance. To prevent both overfitting and distributional shift, we introduce data-regularised environment design (D
  arXiv  Detail & Related papers  (2024-02-05T19:47:45Z)
• Value Explicit Pretraining for Learning Transferable Representations [11.069853883599102]
  We propose a method that learns generalizable representations for transfer reinforcement learning. We learn new tasks that share similar objectives as previously learned tasks, by learning an encoder for objective-conditioned representations. Experiments using a realistic navigation simulator and Atari benchmark show that the pretrained encoder produced by our method outperforms current SoTA pretraining methods.
  arXiv  Detail & Related papers  (2023-12-19T17:12:35Z)
• What Makes Pre-Trained Visual Representations Successful for Robust Manipulation? [57.92924256181857]
  We find that visual representations designed for manipulation and control tasks do not necessarily generalize under subtle changes in lighting and scene texture. We find that emergent segmentation ability is a strong predictor of out-of-distribution generalization among ViT models.
  arXiv  Detail & Related papers  (2023-11-03T18:09:08Z)
• SpawnNet: Learning Generalizable Visuomotor Skills from Pre-trained Networks [52.766795949716986]
  We present a study of the generalization capabilities of the pre-trained visual representations at the categorical level. We propose SpawnNet, a novel two-stream architecture that learns to fuse pre-trained multi-layer representations into a separate network to learn a robust policy.
  arXiv  Detail & Related papers  (2023-07-07T13:01:29Z)
• ALP: Action-Aware Embodied Learning for Perception [60.64801970249279]
  We introduce Action-Aware Embodied Learning for Perception (ALP) ALP incorporates action information into representation learning through a combination of optimizing a reinforcement learning policy and an inverse dynamics prediction objective. We show that ALP outperforms existing baselines in several downstream perception tasks.
  arXiv  Detail & Related papers  (2023-06-16T21:51:04Z)
• Improving Meta-Learning Generalization with Activation-Based Early-Stopping [12.299371455015239]
  Meta-Learning algorithms for few-shot learning aim to train neural networks capable of generalizing to novel tasks using only a few examples. Early-stopping is critical for performance, halting model training when it reaches optimal generalization to the new task distribution. This is problematic in few-shot transfer learning settings, where the meta-test set comes from a different target dataset.
  arXiv  Detail & Related papers  (2022-08-03T22:55:45Z)
• Unifying Language Learning Paradigms [96.35981503087567]
  We present a unified framework for pre-training models that are universally effective across datasets and setups. We show how different pre-training objectives can be cast as one another and how interpolating between different objectives can be effective. Our model also achieve strong results at in-context learning, outperforming 175B GPT-3 on zero-shot SuperGLUE and tripling the performance of T5-XXL on one-shot summarization.
  arXiv  Detail & Related papers  (2022-05-10T19:32:20Z)
• Instance based Generalization in Reinforcement Learning [24.485597364200824]
  We analyze policy learning in the context of Partially Observable Markov Decision Processes (POMDPs) We prove that, independently of the exploration strategy, reusing instances introduces significant changes on the effective Markov dynamics the agent observes during training. We propose training a shared belief representation over an ensemble of specialized policies, from which we compute a consensus policy that is used for data collection, disallowing instance specific exploitation.
  arXiv  Detail & Related papers  (2020-11-02T16:19:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.