Paper improves Thompson Sampling for linear contextual bandits.
problem Empirical Thompson Sampling does not achieve optimal regret bounds.
method Develops a novel estimator with adaptive data augmentation and coupling.
result Achieves nearly minimax optimal performance.
The paper uses potential functions to help reinforcement learning agents learn optimal policies.
problem Learning optimal stochastic policies in reinforcement learning.
method Augmenting the reward with potential functions and applying these to policy gradient algorithms.
result Potential-based reward shaping preserves optimality of stochastic policies and speeds up learning.
New method learns various data manipulation schemes for model training.
problem Improving model training with data manipulation.
method Adapts RL reward learning algorithm for data manipulation learning.
result Significant improvement in classification performance.
Paper introduces PRMs to learn non-Markovian stochastic rewards for reinforcement learning.
problem Lack of structured representation for non-Markovian stochastic rewards in reinforcement learning.
method Introduces probabilistic reward machines (PRMs) and presents an algorithm to learn them from decision processes.
result Algorithm proves correct and convergent for learning PRMs from decision processes.
Capsule Networks improve autonomous navigation in sparse environments.
problem Challenges in reinforcement learning for sparse reward environments.
method Caps-EM pairs CapsNets with Advantage Actor Critic, using fewer parameters.
result Caps-EM achieves significant time improvements and fewer parameters compared to competing methods.
Self-play is an unsupervised training procedure which enables the reinforcement learning agents to explore the environment without requiring any external rewards. We augment the self-play setting by providing an external memory where the agent can store experience from the previous tasks. This enables the agent to come…
Enhances reinforcement learning from sparse data.
problem Limited data for offline reinforcement learning.
method Trajectory-based data augmentation.
result Improves reinforcement learning performance.
We address the problem of regret minimization in logistic contextual bandits, where a learner decides among sequential actions or arms given their respective contexts to maximize binary rewards. Using a fast inference procedure with Polya-Gamma distributed augmentation variables, we propose an improved version of Thomp…
Self-supervised reward prediction improves RL in sparse reward settings.
problem Data efficiency and sparse reward signals in reinforcement learning.
method Learning a state representation for reward prediction and using it to shape rewards.
result Self-supervised reward prediction enhances RL algorithms in single-goal environments.
A new method resolves non-identifiability in reward modeling using anchor labels.
problem Non-identifiability in reward modeling from pairwise preferences alone.
method Anchor-guided Variance-aware Reward Modeling (AVRM) framework.
result AVRM resolves non-identifiability and improves reward modeling performance.
This paper improves auto-augment efficiency by sharing augmentation weights.
problem Efficient evaluation of augmentation policies for model training.
method Augmentation-Wise Weight Sharing (AWS) to create a fast yet accurate proxy task.
result Augmentation policies found achieve superior accuracies compared to existing methods.
Improved reinforcement learning with deep learning.
problem Extending MultiGrid Reinforcement Learning to work with deep learning.
method Combining potential-based reward shaping with a learned potential function from interaction, and adapting it for deep learning algorithms.
result DQN augmented with the approach performs significantly better on continuous control tasks.
Paper tackles robust offline RL with heavy-tailed rewards.
problem Real-world applications often encounter heavy-tailed rewards, challenging offline RL.
method Proposes ROAM and ROOM algorithms using median-of-means method for robust off-policy evaluation and OPO.
result Demonstrates superior performance on heavy-tailed reward datasets compared to existing methods.
Trade-R1 bridges verifiable rewards to stochastic financial markets via process-level reasoning verification.
problem Extending RL to financial markets where rewards are verifiable but noisy.
method A verification method that transforms reasoning over financial documents into a structured RAG task, using a triangular consistency metric.
result DSR achieves superior cross-market generalization while maintaining reasoning consistency.
In many sequential decision making tasks, it is challenging to design reward functions that help an RL agent efficiently learn behavior that is considered good by the agent designer. A number of different formulations of the reward-design problem, or close variants thereof, have been proposed in the literature. In this…
Paper estimates risks in MDPs using state lumping and SAT, showing its effectiveness.
problem Estimating risks in Markov decision processes with state augmentation.
method State augmentation transformation, isotopic states, and state lumping.
result SAT and state lumping effectively estimate mean-variance and exponential utility risks.
New models for bandit problems with fidelity rewards are introduced and analyzed.
problem Fidelity rewards in bandit problems to incentivize loyalty.
method Two models (loyalty-points and subscription) for fidelity rewards; stochastic and adversarial settings considered.
result Sublinear regret bounds for some models, worst case lower bounds for others.
Improves generative models by optimizing rewards and sample editing.
problem Efficiently generating high-reward samples with structural constraints.
method Introduces MDM-VGB, a discrete diffusion sampler that augments unmasking generation with reward-guided remasking.
result MDM-VGB achieves quadratic complexity and robustness to noise, outperforming heuristics like best-of-N. We present Memory Augmented Policy Optimization (MAPO), a simple and novel way to leverage a memory buffer of promising trajectories to reduce the variance of policy gradient estimate. MAPO is applicable to deterministic environments with discrete actions, such as structured prediction and combinatorial optimization ta…
IPO optimizes reinforcement learning with constraints for better performance.
problem Maximizing long-term reward while satisfying cumulative constraints in decision problems.
method Interior-point Policy Optimization (IPO) using logarithmic barrier functions.
result IPO outperforms state-of-the-art baselines in reward maximization and constraint satisfaction.
We propose an expert-augmented actor-critic algorithm, which we evaluate on two environments with sparse rewards: Montezumas Revenge and a demanding maze from the ViZDoom suite. In the case of Montezumas Revenge, an agent trained with our method achieves very good results consistently scoring above 27,000 points (in ma…
RePULSe improves language model alignment by reducing undesired outputs without sacrificing overall performance.
problem Aligning language models with human preferences while minimizing undesired outputs.
method Integrates probabilistic inference into RL training to reduce undesired outputs.
result RePULSe achieves a better balance between expected reward and undesired output probability.
Deep learning has achieved remarkable successes in solving challenging reinforcement learning (RL) problems when dense reward function is provided. However, in sparse reward environment it still often suffers from the need to carefully shape reward function to guide policy optimization. This limits the applicability of…
Entropy regularization improves policy optimization in reinforcement learning.
problem Improving policy optimization in reinforcement learning.
method Entropy regularization is introduced to soften the greedy policy towards a more diverse softmax policy, leading to a continuously parameterized algorithm that interpolates between policy gradient and Q-learning.
result An intermediate algorithm can improve performance in reinforcement learning.
New algorithms detect changes in non-stationary MABs for better performance.
problem Non-stationary MAB environments where arm reward distributions change over time.
method Modular Detection Augmented Bandit (DAB) procedures with improved performance lower bounds.
result Modular DAB procedures achieve order-optimal regret bounds for various change detectors and bandit algorithms.
The paper enhances preference learning by incorporating response time data.
problem Lack of temporal information in user decision-making for reward model learning.
method Integrates response time alongside binary choice data using the EZ model and Neyman-orthogonal loss functions.
result Response time-augmented approach reduces error rates from exponential to polynomial scaling, improving sample efficiency.
Improved AST method finds more useful failure scenarios for autonomous vehicles.
problem Finding useful failure scenarios for autonomous vehicle validation is challenging.
method Adaptive Stress Testing with reward augmentation, modified to encode domain information.
result The modified AST method discovers a larger and more expressive subset of failure scenarios.
Deep Reinforcement Learning has been shown to be very successful in complex games, e.g. Atari or Go. These games have clearly defined rules, and hence allow simulation. In many practical applications, however, interactions with the environment are costly and a good simulator of the environment is not available. Further…
Active inference enhances RL by balancing exploration and exploitation.
problem Traditional RL's balance between exploration and exploitation is often suboptimal.
method Developed a new decision-making objective based on active inference.
result The new algorithm successfully balances exploration and exploitation on various RL benchmarks.
Training-free method improves large language model sequence quality via reward-guided sampling.
problem Optimizing large language model sequence quality over token likelihood.
method Reward-augmented target distribution combined with Sequential Monte Carlo sampling.
result Significant gains in sequence generation and mathematical reasoning tasks.
New unsupervised learning task improves RL performance.
problem Reward-driven feature learning limitations in RL from images.
method Introduce Augmented Temporal Contrast (ATC) for unsupervised learning of image representations.
result Training encoders using ATC matches or outperforms end-to-end RL in most environments.
A new algorithm STE for model-based RL improves learning rates.
problem Sparse rewards and computational intractability of estimating information gain.
method Developed a novel algorithm based on Stein Information Directed Exploration (STE)E.
result Achieves sublinear Bayesian regret, outperforming prior approaches.
Enhances RL in target domains with limited data using augmented return.
problem Utilize data from an accessible source domain to improve policy learning in a target domain with scarce data.
method Return Augmented Decision Transformer (REAG) method, which augments the return in the source domain to align with the target domain's optimal trajectory distribution.
result The proposed REAG method achieves the same level of suboptimality as without a dynamics shift, enhancing DT type frameworks' performance in off-dynamics RL.
Task-agnostic RL tackles exploration in MDPs with multiple tasks.
problem Challenges in reinforcement learning with multiple tasks or conflicting objectives.
method Task-agnostic RL framework, UCBZero algorithm.
result UCBZero finds near-optimal policies for multiple tasks efficiently.
We design and study a Contextual Memory Tree (CMT), a learning memory controller that inserts new memories into an experience store of unbounded size. It is designed to efficiently query for memories from that store, supporting logarithmic time insertion and retrieval operations. Hence CMT can be integrated into existi…
A number of recent approaches to policy learning in 2D game domains have been successful going directly from raw input images to actions. However when employed in complex 3D environments, they typically suffer from challenges related to partial observability, combinatorial exploration spaces, path planning, and a scarc…
SAVED safely learns robot tasks with sparse rewards.
problem Challenges in reinforcement learning for robotics, especially sparse rewards and complex constraints.
method SAVED uses supervision to constrain exploration and learn efficiently, handling complex constraints.
result SAVED outperforms state-of-the-art methods in success rate, constraint satisfaction, and sample efficiency.
A new method shapes reinforcement learning environments by abstracting large state spaces.
problem Learning in large, noisy environments with sparse feedback.
method Environment shaping using state abstraction.
result Agent's policy in shaped environment preserves near-optimal behavior in original environment.
CCLF improves sample efficiency in RL by prioritizing informative samples.
problem Learning from high-dimensional observations is challenging and unstable.
method Contrastive-Curiosity-Driven Learning Framework (CCLF) that prioritizes informative samples.
result CCLF improves sample efficiency and learning performance in RL.
Model user preferences for conversational LLMs using weak rewards.
problem Lack of persistent user models in conversational LLMs leading to repeated user restatements.
method Vector-Adapted Retrieval Scoring (VARS) framework that updates user vectors online from weak scalar rewards.
result Full VARS agent achieves strongest overall performance, matches strong Reflection baseline in task success, and reduces user effort.
Causal KL improves on existing metrics for evaluating causal models.
problem Insufficient discrimination between causal models using edit-distance and KL divergence.
method Introducing Causal KL, an augmented KL divergence that considers causal relationships.
result Causal KL variants effectively distinguish between observationally equivalent models.
In this work, we study the credit assignment problem in reward augmented maximum likelihood (RAML) learning, and establish a theoretical equivalence between the token-level counterpart of RAML and the entropy regularized reinforcement learning. Inspired by the connection, we propose two sequence prediction algorithms, …
DSAC improves cooperative MARL with general utilities, converging faster than existing methods.
problem Improving cooperation in multi-agent reinforcement learning with nonlinear utilities.
method Decentralized Shadow Reward Actor-Critic (DSAC) that estimates local occupancy measures and derivatives.
result DSAC converges to ε-stationarity in O(1/ε^2.5) steps with high probability, finding globally optimal policies.
Recent work has identified that classification models implemented as neural networks are vulnerable to data-poisoning and Trojan attacks at training time. In this work, we show that these training-time vulnerabilities extend to deep reinforcement learning (DRL) agents and can be exploited by an adversary with access to…
Reward augmented maximum likelihood (RAML), a simple and effective learning framework to directly optimize towards the reward function in structured prediction tasks, has led to a number of impressive empirical successes. RAML incorporates task-specific reward by performing maximum-likelihood updates on candidate outpu…
RL agents learn from a few tasks to generalize to new ones.
problem Creating efficient RL agents that can solve multiple tasks.
method GHP-MDPs model with latent variables for hidden parameters.
result State-of-the-art performance and sample-efficiency on new tasks.
Paper tackles reinforcement learning for STL specifications with state history.
problem Learning optimal policies to satisfy STL specifications often requires too much state history, making the problem computationally intractable.
method Proposes a compact augmented state-space representation to capture state history and an approximation method to solve the objective.
result Shows the performance bound of the approximate solution and compares it with an existing technique.
BYOL-Explore learns to explore visually-rich environments by predicting world dynamics.
problem Exploration in visually complex environments.
method Optimizes a single prediction loss in latent space to learn world representation, dynamics, and exploration policy.
result Achieves superhuman performance on Atari games with simpler design.