Transfer learning in Reinforcement Learning (RL) enables agents to leverage knowledge from source tasks to accelerate learning in target tasks. While prior work, such as the Attend, Adapt, and Transfer (A2T) framework, addresses negative transfer and selective transfer, other critical challenges remain underexplored. This paper introduces the Generalized Adaptive Transfer Network (GATN), a deep RL architecture designed to tackle task generalization across domains, robustness to environmental changes, and computational efficiency in transfer. GATN employs a domain-agnostic representation module, a robustness-aware policy adapter, and an efficient transfer scheduler to achieve these goals. We evaluate GATN on diverse benchmarks, including Atari 2600, MuJoCo, and a custom chatbot dialogue environment, demonstrating superior performance in cross-domain generalization, resilience to dynamic environments, and reduced computational overhead compared to baselines. Our findings suggest GATN is a versatile framework for real-world RL applications, such as adaptive chatbots and robotic control.
Deep Reinforcement Learning (DRL) has achieved remarkable success in complex sequential decision-making tasks, such as playing Atari 2600 games and mastering board games. A critical yet underexplored aspect of DRL is the temporal scale of action execution. We propose a novel paradigm that integrates contextual bandits with DRL to adaptively select action durations, enhancing policy flexibility and computational efficiency. Our approach augments a Deep Q-Network (DQN) with a contextual bandit module that learns to choose optimal action repetition rates based on state contexts. Experiments on Atari 2600 games demonstrate significant performance improvements over static duration baselines, highlighting the efficacy of adaptive temporal abstractions in DRL. This paradigm offers a scalable solution for real-time applications like gaming and robotics, where dynamic action durations are critical.
One proposed mechanism to improve exploration in reinforcement learning is through the use of macro-actions. Paradoxically though, in many scenarios the naive addition of macro-actions does not lead to better exploration, but rather the opposite. It has been argued that this was caused by adding non-useful macros and multiple works have focused on mechanisms to discover effectively environment-specific useful macros. In this work, we take a slightly different perspective. We argue that the difficulty stems from the trade-offs between reducing the average number of decisions per episode versus increasing the size of the action space. Namely, one typically treats each potential macro-action as independent and atomic, hence strictly increasing the search space and making typical exploration strategies inefficient. To address this problem we propose a novel regularization term that exploits the relationship between actions and macro-actions to improve the credit assignment mechanism by reducing the effective dimension of the action space and, therefore, improving exploration. The term relies on a similarity matrix that is meta-learned jointly with learning the desired policy. We empirically validate our strategy looking at macro-actions in Atari games, and the StreetFighter II environment. Our results show significant improvements over the Rainbow-DQN baseline in all environments. Additionally, we show that the macro-action similarity is transferable to related environments. We believe this work is a small but important step towards understanding how the similarity-imposed geometry on the action space can be exploited to improve credit assignment and exploration, therefore making learning more effective.
In reinforcement learning, distributional off-policy evaluation (OPE) focuses on estimating the return distribution of a target policy using offline data collected under a different policy. This work focuses on extending the widely used fitted-Q evaluation -- developed for expectation-based reinforcement learning -- to the distributional OPE setting. We refer to this extension as fitted distributional evaluation (FDE). While only a few related approaches exist, there remains no unified framework for designing FDE methods. To fill this gap, we present a set of guiding principles for constructing theoretically grounded FDE methods. Building on these principles, we develop several new FDE methods with convergence analysis and provide theoretical justification for existing methods, even in non-tabular environments. Extensive experiments, including simulations on linear quadratic regulators and Atari games, demonstrate the superior performance of the FDE methods.
We propose a new ternary spiking neuron model to improve the representation capacity of binary spiking neurons in deep Q-learning. Although a ternary neuron model has recently been introduced to overcome the limited representation capacity offered by the binary spiking neurons, we show that its performance is worse than that of binary models in deep Q-learning tasks. We hypothesize gradient estimation bias during the training process as the underlying potential cause through mathematical and empirical analysis. We propose a novel ternary spiking neuron model to mitigate this issue by reducing the estimation bias. We use the proposed ternary spiking neuron as the fundamental computing unit in a deep spiking Q-learning network (DSQN) and evaluate the network's performance in seven Atari games from the Gym environment. Results show that the proposed ternary spiking neuron mitigates the drastic performance degradation of ternary neurons in Q-learning tasks and improves the network performance compared to the existing binary neurons, making DSQN a more practical solution for on-board autonomous decision-making tasks.
Plasticity loss, a critical challenge in neural network training, limits a model's ability to adapt to new tasks or shifts in data distribution. This paper introduces AID (Activation by Interval-wise Dropout), a novel method inspired by Dropout, designed to address plasticity loss. Unlike Dropout, AID generates subnetworks by applying Dropout with different probabilities on each preactivation interval. Theoretical analysis reveals that AID regularizes the network, promoting behavior analogous to that of deep linear networks, which do not suffer from plasticity loss. We validate the effectiveness of AID in maintaining plasticity across various benchmarks, including continual learning tasks on standard image classification datasets such as CIFAR10, CIFAR100, and TinyImageNet. Furthermore, we show that AID enhances reinforcement learning performance in the Arcade Learning Environment benchmark.
We present TextAtari, a benchmark for evaluating language agents on very long-horizon decision-making tasks spanning up to 100,000 steps. By translating the visual state representations of classic Atari games into rich textual descriptions, TextAtari creates a challenging test bed that bridges sequential decision-making with natural language processing. The benchmark includes nearly 100 distinct tasks with varying complexity, action spaces, and planning horizons, all rendered as text through an unsupervised representation learning framework (AtariARI). We evaluate three open-source large language models (Qwen2.5-7B, Gemma-7B, and Llama3.1-8B) across three agent frameworks (zero-shot, few-shot chain-of-thought, and reflection reasoning) to assess how different forms of prior knowledge affect performance on these long-horizon challenges. Four scenarios-Basic, Obscured, Manual Augmentation, and Reference-based-investigate the impact of semantic understanding, instruction comprehension, and expert demonstrations on agent decision-making. Our results reveal significant performance gaps between language agents and human players in extensive planning tasks, highlighting challenges in sequential reasoning, state tracking, and strategic planning across tens of thousands of steps. TextAtari provides standardized evaluation protocols, baseline implementations, and a framework for advancing research at the intersection of language models and planning.
A key task in Artificial Intelligence is learning effective policies for controlling agents in unknown environments to optimize performance measures. Off-policy learning methods, like Q-learning, allow learners to make optimal decisions based on past experiences. This paper studies off-policy learning from biased data in complex and high-dimensional domains where \emph{unobserved confounding} cannot be ruled out a priori. Building on the well-celebrated Deep Q-Network (DQN), we propose a novel deep reinforcement learning algorithm robust to confounding biases in observed data. Specifically, our algorithm attempts to find a safe policy for the worst-case environment compatible with the observations. We apply our method to twelve confounded Atari games, and find that it consistently dominates the standard DQN in all games where the observed input to the behavioral and target policies mismatch and unobserved confounders exist.
One of the primary aspirations in reinforcement learning research is developing general-purpose agents capable of rapidly adapting to and mastering novel tasks. While RL gaming agents have mastered many Atari games, they remain slow and costly to train for each game. In this work, we demonstrate that latest reasoning LLMs with out-of-domain RL post-training can play a challenging Atari game called Frogger under a zero-shot setting. We then investigate the effect of in-context learning and the amount of reasoning effort on LLM performance. Lastly, we demonstrate a way to bootstrap traditional RL method with LLM demonstrations, which significantly improves their performance and sample efficiency. Our implementation is open sourced at https://github.com/AlienKevin/frogger.
Recently, deep reinforcement learning has emerged as a promising approach for solving complex combinatorial optimization problems. Broadly, deep reinforcement learning methods fall into two categories: policy-based and value-based. While value-based approaches have achieved notable success in domains such as the Arcade Learning Environment, the combinatorial optimization community has predominantly favored policy-based methods, often overlooking the potential of value-based algorithms. In this work, we conduct a comprehensive empirical evaluation of value-based algorithms, including the deep q-network and several of its advanced extensions, within the context of two complex combinatorial problems: the job-shop and the flexible job-shop scheduling problems, two fundamental challenges with multiple industrial applications. Our results challenge the assumption that policy-based methods are inherently superior for combinatorial optimization. We show that several value-based approaches can match or even outperform the widely adopted proximal policy optimization algorithm, suggesting that value-based strategies deserve greater attention from the combinatorial optimization community. Our code is openly available at: https://github.com/AJ-Correa/Unraveling-the-Rainbow.
The ability to train intelligent autonomous agents (such as mobile robots) on multiple tasks is crucial for adapting to dynamic real-world environments. However, state-of-the-art reinforcement learning (RL) methods only excel in single-task settings, and still struggle to generalize across multiple tasks due to task interference. Moreover, real-world environments also demand the agents to have data stream processing capabilities. Toward this, a state-of-the-art work employs Spiking Neural Networks (SNNs) to improve multi-task learning by exploiting temporal information in data stream, while enabling lowpower/energy event-based operations. However, it relies on fixed context/task-switching intervals during its training, hence limiting the scalability and effectiveness of multi-task learning. To address these limitations, we propose SwitchMT, a novel adaptive task-switching methodology for RL-based multi-task learning in autonomous agents. Specifically, SwitchMT employs the following key ideas: (1) a Deep Spiking Q-Network with active dendrites and dueling structure, that utilizes task-specific context signals to create specialized sub-networks; and (2) an adaptive task-switching policy that leverages both rewards and internal dynamics of the network parameters. Experimental results demonstrate that SwitchMT achieves superior performance in multi-task learning compared to state-of-the-art methods. It achieves competitive scores in multiple Atari games (i.e., Pong: -8.8, Breakout: 5.6, and Enduro: 355.2) compared to the state-of-the-art, showing its better generalized learning capability. These results highlight the effectiveness of our SwitchMT methodology in addressing task interference while enabling multi-task learning automation through adaptive task switching, thereby paving the way for more efficient generalist agents with scalable multi-task learning capabilities.
Current approaches in Explainable Deep Reinforcement Learning have limitations in which the attention mask has a displacement with the objects in visual input. This work addresses a spatial problem within traditional Convolutional Neural Networks (CNNs). We propose the Interpretable Feature Extractor (IFE) architecture, aimed at generating an accurate attention mask to illustrate both "what" and "where" the agent concentrates on in the spatial domain. Our design incorporates a Human-Understandable Encoding module to generate a fully interpretable attention mask, followed by an Agent-Friendly Encoding module to enhance the agent's learning efficiency. These two components together form the Interpretable Feature Extractor for vision-based deep reinforcement learning to enable the model's interpretability. The resulting attention mask is consistent, highly understandable by humans, accurate in spatial dimension, and effectively highlights important objects or locations in visual input. The Interpretable Feature Extractor is integrated into the Fast and Data-efficient Rainbow framework, and evaluated on 57 ATARI games to show the effectiveness of the proposed approach on Spatial Preservation, Interpretability, and Data-efficiency. Finally, we showcase the versatility of our approach by incorporating the IFE into the Asynchronous Advantage Actor-Critic Model.
Model-based reinforcement learning (MBRL) has demonstrated superior sample efficiency compared to model-free reinforcement learning (MFRL). However, the presence of inaccurate models can introduce biases during policy learning, resulting in misleading trajectories. The challenge lies in obtaining accurate models due to limited diverse training data, particularly in regions with limited visits (uncertain regions). Existing approaches passively quantify uncertainty after sample generation, failing to actively collect uncertain samples that could enhance state coverage and improve model accuracy. Moreover, MBRL often faces difficulties in making accurate multi-step predictions, thereby impacting overall performance. To address these limitations, we propose a novel framework for uncertainty-aware policy optimization with model-based exploratory planning. In the model-based planning phase, we introduce an uncertainty-aware k-step lookahead planning approach to guide action selection at each step. This process involves a trade-off analysis between model uncertainty and value function approximation error, effectively enhancing policy performance. In the policy optimization phase, we leverage an uncertainty-driven exploratory policy to actively collect diverse training samples, resulting in improved model accuracy and overall performance of the RL agent. Our approach offers flexibility and applicability to tasks with varying state/action spaces and reward structures. We validate its effectiveness through experiments on challenging robotic manipulation tasks and Atari games, surpassing state-of-the-art methods with fewer interactions, thereby leading to significant performance improvements.
Recent developments in deep reinforcement learning have been very successful in learning complex, previously intractable problems. Sample efficiency and local optimality, however, remain significant challenges. To address these challenges, novelty-driven exploration strategies have emerged and shown promising potential. Unfortunately, no single algorithm outperforms all others in all tasks and most of them struggle with tasks with high-dimensional and complex observations. In this work, we propose Adventurer, a novelty-driven exploration algorithm that is based on Bidirectional Generative Adversarial Networks (BiGAN), where BiGAN is trained to estimate state novelty. Intuitively, a generator that has been trained on the distribution of visited states should only be able to generate a state coming from the distribution of visited states. As a result, novel states using the generator to reconstruct input states from certain latent representations would lead to larger reconstruction errors. We show that BiGAN performs well in estimating state novelty for complex observations. This novelty estimation method can be combined with intrinsic-reward-based exploration. Our empirical results show that Adventurer produces competitive results on a range of popular benchmark tasks, including continuous robotic manipulation tasks (e.g. Mujoco robotics) and high-dimensional image-based tasks (e.g. Atari games).
Studies in reward shaping for reinforcement learning (RL) have flourished in recent years due to its ability to speed up training. Our previous work proposed an adaptive potential function (APF) and showed that APF can accelerate the Q-learning with a Multi-layer Perceptron algorithm in the low-dimensional domain. This paper proposes to extend APF with an encoder (APF+) for RL state representation, allowing applying APF to the pixel-based Atari games using a state-encoding method that projects high-dimensional game's pixel frames to low-dimensional embeddings. We approach by designing the state-representation encoder as a W-shaped network (W-Net), by using which we are able to encode both the background as well as the moving entities in the game frames. Specifically, the embeddings derived from the pre-trained W-Net consist of two latent vectors: One represents the input state, and the other represents the deviation of the input state's representation from itself. We then incorporate W-Net into APF to train a downstream Dueling Deep Q-Network (DDQN), obtain the APF-WNet-DDQN, and demonstrate its effectiveness in Atari game-playing tasks. To evaluate the APF+W-Net module in such high-dimensional tasks, we compare with two types of baseline methods: (i) the basic DDQN; and (ii) two encoder-replaced APF-DDQN methods where we replace W-Net by (a) an unsupervised state representation method called Spatiotemporal Deep Infomax (ST-DIM) and (b) a ground truth state representation provided by the Atari Annotated RAM Interface (ARI). The experiment results show that out of 20 Atari games, APF-WNet-DDQN outperforms DDQN (14/20 games) and APF-STDIM-DDQN (13/20 games) significantly. In comparison against the APF-ARI-DDQN which employs embeddings directly of the detailed game-internal state information, the APF-WNet-DDQN achieves a comparable performance.
Achieving autonomous agents with robust generalization capabilities across diverse games and tasks remains one of the ultimate goals in AI research. Recent advancements in transformer-based offline reinforcement learning, exemplified by the MultiGame Decision Transformer [Lee et al., 2022], have shown remarkable performance across various games or tasks. However, these approaches depend heavily on human expertise, presenting substantial challenges for practical deployment, particularly in scenarios with limited prior game-specific knowledge. In this paper, we propose an algorithm called Multi-Game Target Return Optimizer (MTRO) to autonomously determine game-specific target returns within the Multi-Game Decision Transformer framework using solely offline datasets. MTRO addresses the existing limitations by automating the target return configuration process, leveraging environmental reward information extracted from offline datasets. Notably, MTRO does not require additional training, enabling seamless integration into existing Multi-Game Decision Transformer architectures. Our experimental evaluations on Atari games demonstrate that MTRO enhances the performance of RL policies across a wide array of games, underscoring its potential to advance the field of autonomous agent development.
Planning with options -- a sequence of primitive actions -- has been shown effective in reinforcement learning within complex environments. Previous studies have focused on planning with predefined options or learned options through expert demonstration data. Inspired by MuZero, which learns superhuman heuristics without any human knowledge, we propose a novel approach, named OptionZero. OptionZero incorporates an option network into MuZero, providing autonomous discovery of options through self-play games. Furthermore, we modify the dynamics network to provide environment transitions when using options, allowing searching deeper under the same simulation constraints. Empirical experiments conducted in 26 Atari games demonstrate that OptionZero outperforms MuZero, achieving a 131.58% improvement in mean human-normalized score. Our behavior analysis shows that OptionZero not only learns options but also acquires strategic skills tailored to different game characteristics. Our findings show promising directions for discovering and using options in planning. Our code is available at https://rlg.iis.sinica.edu.tw/papers/optionzero.
Autonomous artificial agents must be able to learn behaviors in complex environments without humans to design tasks and rewards. Designing these functions for each environment is not feasible, thus, motivating the development of intrinsic reward functions. In this paper, we propose using several cognitive elements that have been neglected for a long time to build an internal world model for an intrinsically motivated agent. Our agent performs satisfactory iterations with the environment, learning complex behaviors without needing previously designed reward functions. We used 18 Atari games to evaluate what cognitive skills emerge in games that require reactive and deliberative behaviors. Our results show superior performance compared to the state-of-the-art in many test cases with dense and sparse rewards.
World model (WM) agents enable sample-efficient reinforcement learning by learning policies entirely from simulated experience. However, existing token-based world models (TBWMs) are limited to visual inputs and discrete actions, restricting their adoption and applicability. Moreover, although both intrinsic motivation and prioritized WM replay have shown promise in improving WM performance and generalization, they remain underexplored in this setting, particularly in combination. We introduce Simulus, a highly modular TBWM agent that integrates (1) a modular multi-modality tokenization framework, (2) intrinsic motivation, (3) prioritized WM replay, and (4) regression-as-classification for reward and return prediction. Simulus achieves state-of-the-art sample efficiency for planning-free WMs across three diverse benchmarks. Ablation studies reveal the individual contribution of each component while highlighting their synergy. Our code and model weights are publicly available at https://github.com/leor-c/Simulus.
Reinforcement Learning (RL) has been widely used in many applications, particularly in gaming, which serves as an excellent training ground for AI models. Google DeepMind has pioneered innovations in this field, employing reinforcement learning algorithms, including model-based, model-free, and deep Q-network approaches, to create advanced AI models such as AlphaGo, AlphaGo Zero, and MuZero. AlphaGo, the initial model, integrates supervised learning and reinforcement learning to master the game of Go, surpassing professional human players. AlphaGo Zero refines this approach by eliminating reliance on human gameplay data, instead utilizing self-play for enhanced learning efficiency. MuZero further extends these advancements by learning the underlying dynamics of game environments without explicit knowledge of the rules, achieving adaptability across various games, including complex Atari games. This paper reviews the significance of reinforcement learning applications in Atari and strategy-based games, analyzing these three models, their key innovations, training processes, challenges encountered, and improvements made. Additionally, we discuss advancements in the field of gaming, including MiniZero and multi-agent models, highlighting future directions and emerging AI models from Google DeepMind.