BASIS improves LLM reasoning by sharing batchwise rollout info, reducing MSE by 69%.
problem Improving large language model reasoning with limited rollouts and batch information.
method BASIS samples only one rollout per prompt but uses batch information to improve value function estimation.
result BASIS reduces MSE in value function estimation by 69% compared to REINFORCE++.
Efficiently reduces computational burden of rollout acquisition functions in Bayesian optimization.
problem Expensive computation of rollout acquisition functions in Bayesian optimization.
method Combines quasi-Monte Carlo, common random numbers, and control variates to reduce computational burden. Formulates a policy-search approach to eliminate the need to optimize the rollout acquisition function.
result Significant reduction in computational burden of rollout acquisition functions.
Deep neural networks, and in particular recurrent networks, are promising candidates to control autonomous agents that interact in real-time with the physical world. However, this requires a seamless integration of temporal features into the network's architecture. For the training of and inference with recurrent neura…
A new estimator combines bootstrapping and rollout methods in RL.
problem Combining strengths of bootstrapping and rollout methods in RL.
method Subgraph Bellman operators and fixed point solving.
result Upper bound on error approaches optimal TD variance with additional term.
Paper proposes a forecasting solution for network-rollout planning.
problem Accurate estimation of milestone completion times for network-rollout planning.
method Partition-based regression models incorporating data-driven statistical models.
result The proposed approach outperforms alternative models in terms of performance and model complexity.
Sequence generation models are commonly refined with reinforcement learning over user-defined metrics. However, high gradient variance hinders the practical use of this method. To stabilize this method, we adapt to contextual generation of categorical sequences a policy gradient estimator, which evaluates a set of corr…
A new method reduces compounding errors in model-based reinforcement learning.
problem Compounding errors in long horizon predictions from model-based reinforcement learning.
method Maximum Entropy Model Rollouts (MEMR) with non-uniform sampling and prioritized experience replay.
result Significantly reduces computation requirements compared to other model-based methods.
Lookahead, also known as non-myopic, Bayesian optimization (BO) aims to find optimal sampling policies through solving a dynamic program (DP) that maximizes a long-term reward over a rolling horizon. Though promising, lookahead BO faces the risk of error propagation through its increased dependence on a possibly mis-sp…
There are two halves to RL systems: experience collection time and policy learning time. For a large number of samples in rollouts, experience collection time is the major bottleneck. Thus, it is necessary to speed up the rollout generation time with multi-process architecture support. Our work, dubbed WALL-E, utilizes…
AIS corrects rollout-training mismatch in quantized RL, improving speed and stability.
problem Rollout-training mismatch in quantized RL causes bias and training collapse.
method Adaptive Importance Sampling (AIS) adjusts gradient correction per batch.
result AIS matches BF16 baseline on most tasks while improving speed.
Study on indexability of restless multi-armed bandits and rollout policy performance.
problem Maximizing discounted rewards in finite state restless multi-armed bandit problems.
method Decouple the problem into single-armed restless bandits, analyze using value iteration, and compare with Whittle index policy.
result Demonstrates conditions for indexability and compares performance of index policy and rollout policy.
The decision to rollout a vehicle is critical to fleet management companies as wrong decisions can lead to additional cost of maintenance and failures during journey. With the availability of large amount of data and advancement of machine learning techniques, the rollout decisions of a supervisor can be effectively au…
Optimal sample complexity for autoregressive chain-of-thought learning proven.
problem Determining the minimum number of samples needed for accurate autoregressive chain-of-thought learning.
method Proved upper bound on sample complexity using Daniely-Shalev-Shwartz dimension and roll-out stable parity dimension.
result The sample complexity is bounded by the local next-token class rate, with no dependence on rollout length.
SFPO optimizes LLM reasoning by repositioning before updating, improving stability and efficiency.
problem Noisy gradients from low-quality rollouts cause instability and inefficient exploration in on-policy RL algorithms.
method Decomposes each step into three stages: a short fast trajectory, repositioning, and slow correction, preserving the objective and rollout process unchanged.
result SFPO consistently improves stability, reduces rollouts, and accelerates convergence, outperforming GRPO on math reasoning benchmarks.
Model-based reinforcement learning (MBRL) aims to learn a dynamic model to reduce the number of interactions with real-world environments. However, due to estimation error, rollouts in the learned model, especially those of long horizons, fail to match the ones in real-world environments. This mismatching has seriously…
Thermalizer stabilizes autoregressive models for long-term predictions in chaotic systems.
problem Long-term predictions in chaotic spatiotemporal systems are unreliable due to trajectory divergence.
method Diffusion models are used to implicitly estimate the score of an invariant measure, which stabilizes autoregressive emulators by applying denoising during inference.
result Thermalization extends the time horizon of stable predictions by an order of magnitude in chaotic systems.
CoDistill-GRPO improves small models in GRPO by distilling knowledge from a larger model.
problem Small models in GRPO struggle with sparse rewards on difficult tasks.
method Simultaneously trains a large and small model using co-distillation and GRPO objectives.
result Significant improvement in small model performance over standard GRPO on mathematical benchmarks.
Study examines how COVID-19 vaccine companies' popularity affects their stock prices.
problem Impact of COVID-19 vaccine development and rollout on stock prices and company popularity.
method Used Python and various libraries to analyze Google Trends data and stock prices of five vaccine companies.
result Significant correlation between Google Trend data and stock prices, with post-rollout periods showing a slight negative correlation.
New method finds unseen states for RL, improving performance.
problem Offline RL struggles with unseen states and actions.
method Value-informed state perturbations and filtering.
result Improved performance in offline RL tasks.
Theory for RLHF generalization under reward shift and clipped KL.
problem Theoretical understanding of RLHF generalization, especially with reward shift and clipped KL.
method Developed generalization theory for RLHF, accounting for reward shift and clipped KL.
result Presented generalization bounds for RLHF, suggesting generalization error from sampling, reward shift, and KL clipping.
InfoTree improves reinforcement learning by optimizing tool use with a greedy submodular approach.
problem Maximizing information from tool use in reinforcement learning with limited resources.
method Formalizes Rollout Informativeness, recasts state selection as submodular maximization, and uses UUCB and ABA.
result InfoTree outperforms existing methods across various benchmarks, improving performance by 18.2% on average.
This paper introduces a simple, general framework for likelihood-free Bayesian reinforcement learning, through Approximate Bayesian Computation (ABC). The main advantage is that we only require a prior distribution on a class of simulators (generative models). This is useful in domains where an analytical probabilistic…
UK's rapid vaccine rollout linked to reduced COVID-19 mortality.
problem Assessing the impact of accelerated vaccine rollout on public health outcomes.
method Flexible probabilistic models combining interrupted time series analysis and synthetic control methods with multi-output Gaussian processes.
result Substantial reduction in COVID-19 mortality with little effect on transmission rates.
sGPO optimizes RLVR training by balancing inference FLOPs and training efficiency.
problem RLVR training allocates rollout budget without considering query difficulty.
method sGPO trades inference FLOPs for reduced training FLOPs.
result sGPO matches or exceeds baseline performance while reducing training compute by a factor of three.
Several approximate policy iteration schemes without value functions, which focus on policy representation using classifiers and address policy learning as a supervised learning problem, have been proposed recently. Finding good policies with such methods requires not only an appropriate classifier, but also reliable e…
Optimizing news headlines is important for publishers and media sites. A compelling headline will increase readership, user engagement and social shares. At Yahoo Front Page, headline testing is carried out using a test-rollout strategy: we first allocate equal proportion of the traffic to each headline variation for a…
New method optimizes language model performance for test-time strategies.
problem Mismatch between training objectives and test-time deployment of large language models.
method Tail-Extrapolated estimators to approximate best-of-N performance from limited training rollouts.
result Improved performance of best-of-N deployment across various models and datasets.
This paper optimizes sampling policies for Bayesian optimization to improve exploration and exploitation.
problem Improving the balance between exploration and exploitation in Bayesian optimization.
method Developed efficient methods to estimate and optimize non-myopic acquisition functions using rollout policies and stochastic gradient optimization.
result Efficient optimization of sampling policies leads to better performance in Bayesian optimization.
UWM-JEPA predicts future scenarios in belief space, improving accuracy in partially observed environments.
problem Predicting future scenarios in partially observed environments with uncertainty.
method Introduces UWM-JEPA, a JEPA world model with a density-matrix latent and learned unitary predictor.
result UWM-JEPA achieves 0.77 accuracy on a hidden-velocity indicator task, outperforming LSTM-JEPA.
Bayesian inference models failure distributions in autonomous systems.
problem Estimating the distribution of failures in complex systems.
method Bayesian inference using system dynamics rollouts and gradient computation.
result Improves sample efficiency and parameter space coverage in autonomous systems.
Monte Carlo Tree Search (MCTS) algorithms have achieved great success on many challenging benchmarks (e.g., Computer Go). However, they generally require a large number of rollouts, making their applications costly. Furthermore, it is also extremely challenging to parallelize MCTS due to its inherent sequential nature:…
We propose probabilistic models that can extrapolate learning curves of iterative machine learning algorithms, such as stochastic gradient descent for training deep networks, based on training data with variable-length learning curves. We study instantiations of this framework based on random forests and Bayesian recur…
A new method predicts non-Markovian closure terms for complex systems.
problem Predicting the effect of unresolved variables on resolved dynamics in high-dimensional systems.
method Mamba-Assisted Closure (MAC) framework: sequence model trained to predict closure from resolved trajectory, coupled with reduced-order equations.
result Substantially outperforms existing methods in predictive accuracy and long-time stability.
In the past few years, off-policy reinforcement learning methods have shown promising results in their application for robot control. Deep Q-learning, however, still suffers from poor data-efficiency and is susceptible to stochasticity in the environment or reward functions which is limiting with regard to real-world a…
Improves data efficiency in multi-agent control tasks using model-based reinforcement learning.
problem Limited data efficiency in reinforcement learning for multi-agent tasks.
method Decentralized model-based policy optimization (DMPO) framework.
result DMPO achieves superior data efficiency and matches model-free methods using true models.
SAGE enhances reinforcement learning by injecting hints to prevent model stagnation.
problem Sparse rewards cause large language models to stall under relative policy optimization.
method SAGE injects privileged hints during training to increase within-group outcome diversity.
result SAGE consistently outperforms GRPO on 6 benchmarks with LLMs, achieving significant improvements.
Predicts solar dynamics with diffusion models, improving long-range dependencies.
problem Predicting solar dynamics with limited observable data.
method Multiscale inference scheme for diffusion models.
result Improved long-range predictions with reduced bias.
SyMetric evaluates learned Hamiltonian dynamics from images, improving model stability and interpretability.
problem Lack of reliable metrics to assess learned Hamiltonian dynamics from images.
method Developed SyMetric, a binary indicator based on Hamiltonian dynamics properties.
result SyMetric identifies architectural improvements for better dynamics learning.
Generative model learns to mimic time-series behavior to generate accurate trajectories.
problem Learning generative models for time-series data with accurate multi-step trajectories.
method Contrastive imitation framework combining autoregressive and adversarial elements.
result The method generates accurate and useful samples from real-world datasets.
Improved neural PDEs trained on augmented data enhance model accuracy and efficiency.
problem Training neural PDEs on limited data to accurately represent complex systems.
method Space-filling sampling of local states to generate augmented training data.
result Data-augmented neural PDEs outperform traditional emulators in accuracy and stability.
Fine-tuning improves information conveyance in language models by reorganizing uncertainty into more informative sequences.
problem Uncertainty reduction in large language models through fine-tuning is not fully understood, especially regarding output length.
method Proposed Canopy Entropy (CE⋆) to measure uncertainty in both output length and sequence, capturing total Shannon entropy. result Fine-tuned models exhibit stronger positive correlation between entropy rate and semantic diversity, indicating more informative and semantically meaningful generations.
This paper studies effective parallelization of MCTS for computer games.
problem Computational expense of MCTS due to many rollouts.
method Examined parallel MCTS performance, identified necessary conditions for desirable parallelization, and proposed BU-UCT algorithm.
result Theoretical foundation for effective parallel MCTS, with BU-UCT outperforming baselines.
HS-FNO models non-Markovian PDEs by learning history and future states.
problem Non-Markovian dynamics where future states depend on past history.
method History-Space Fourier Neural Operator (HS-FNO) for delay and memory-driven PDEs.
result HS-FNO achieves lowest aggregate errors across various PDE families.
New RL algorithm explains why deep learning works in stochastic environments.
problem Why deep RL algorithms perform well in practice despite using random exploration.
method Introducing SQIRL, an iterative RL algorithm that separates exploration and learning.
result Effective horizon explains why deep RL works in stochastic environments.
Deep reinforcement learning has achieved great successes in recent years, however, one main challenge is the sample inefficiency. In this paper, we focus on how to use action guidance by means of a non-expert demonstrator to improve sample efficiency in a domain with sparse, delayed, and possibly deceptive rewards: the…
This paper proposes a novel deep reinforcement learning architecture that was inspired by previous tree structured architectures which were only useable in discrete action spaces. Policy Prediction Network offers a way to improve sample complexity and performance on continuous control problems in exchange for extra com…
We present an approach to identify concise equations from data using a shallow neural network approach. In contrast to ordinary black-box regression, this approach allows understanding functional relations and generalizing them from observed data to unseen parts of the parameter space. We show how to extend the class o…
CEM-GD combines CEM and gradient descent for efficient model-based RL.
problem Efficient planning in continuous control settings with large prediction horizons.
method Combines CEM for exploration and gradient descent for exploitation.
result Achieves better performance with fewer samples and less computation time.