Action guidance helps agents learn true objectives in games with sparse rewards.
problem Training agents in games with sparse rewards requires significant exploration.
method Action guidance, a novel technique that combines exploration with reward shaping.
result Action guidance enables agents to optimize true objectives efficiently.
This paper proposes a new algorithm for learning guidance rewards in RL.
problem Long-term temporal credit assignment in sparse or delayed reward environments.
method Surrogate RL objective with trajectory-space smoothing to learn guidance rewards.
result Guidance rewards can be learned without additional neural networks and have intuitive interpretation.
Unified framework improves diffusion model rewards without full trajectories.
problem Limited theoretical understanding of guided diffusion samplers.
method Developed a unified algorithmic and theoretical framework for diffusion guidance and reward-guided diffusion.
result Framework shows CFG decreases expected reciprocal of classifier probability.
Novel alignment framework for text-to-image generation using diffusion models and flow matching.
problem Improving text-to-image generation with minimal computational resources.
method Proposes a novel alignment framework that decomposes the score function into pre-trained score plus a conditional expectation of the reward.
result Achieves comparable performance to finetuning-based models with reduced computational cost.
A new method learns continuous guidance weights to improve diffusion model quality and distributional alignment.
problem Improving perceptual quality and distributional alignment of samples from conditional diffusion models.
method Learned continuous guidance weights ωc,(s,t) are used to minimize distributional mismatch and reward guided sampling. result Improvements in Fréchet inception distance (FID) for image generation and better image-prompt alignment in text-to-image applications.
The abstract explores connections between reinforcement learning, scaling, and diffusion.
problem Aligning reinforcement learning with human feedback and scaling techniques.
method Clarifying connections between reinforcement learning, scaling, and diffusion.
result Introducing a resampling approach for alignment and reward-directed diffusion models.
Tutorial on optimizing diffusion model samples for specific metrics.
problem Optimizing diffusion model samples for specific downstream metrics.
method Review and exploration of inference-time guidance and alignment methods.
result Unified perspective on inference-time algorithms and novel methods.
Exploration efficiency is a challenging problem in multi-agent reinforcement learning (MARL), as the policy learned by confederate MARL depends on the collaborative approach among multiple agents. Another important problem is the less informative reward restricts the learning speed of MARL compared with the informative…
New method optimizes diffusion models without fine-tuning, integrating soft value functions.
problem Optimizing natural design spaces of images, molecules, DNA, RNA, and protein sequences.
method Iterative sampling method integrating soft value functions into diffusion model inference.
result Directly utilizes non-differentiable features/reward feedback, applies to discrete diffusion models.
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.
A RL-based method adds conditional controls to pre-trained diffusion models.
problem Precise control over generated samples in diffusion models.
method Formulates the task as an RL problem, using a classifier and KL divergence as reward functions.
result Produces soft-optimal policies that maximize reward functions, enabling conditional sampling.
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…
New approach uses unlabeled prior data to accelerate exploration in sparse reward tasks.
problem Sparse reward tasks in reinforcement learning.
method Learn reward model from online experience, label prior data, and use concurrently.
result Rapid exploration in challenging sparse-reward domains.
New algorithm finds optimal policies without knowing reward functions.
problem Reward-agnostic exploration in reinforcement learning.
method Designs an algorithm that explores without reward information, achieving minimax optimality.
result Achieves provable minimax optimality in finding optimal policies for multiple reward functions.
We propose a framework for ensuring safe behavior of a reinforcement learning agent when the reward function may be difficult to specify. In order to do this, we rely on the existence of demonstrations from expert policies, and we provide a theoretical framework for the agent to optimize in the space of rewards consist…
GSI improves efficiency of large language model inference.
problem Efficiently guiding test-time alignment in large language models.
method Combines soft best-of-n scaling with a reward model and speculative samples. result Achieves higher accuracy and reduced latency compared to standard methods.
Reinforcement Learning in domains with sparse rewards is a difficult problem, and a large part of the training process is often spent searching the state space in a more or less random fashion for any learning signals. For control problems, we often have some controller readily available which might be suboptimal but n…
Data in real-world application often exhibit skewed class distribution which poses an intense challenge for machine learning. Conventional classification algorithms are not effective in the case of imbalanced data distribution, and may fail when the data distribution is highly imbalanced. To address this issue, we prop…
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.
Goals for reinforcement learning problems are typically defined through hand-specified rewards. To design such problems, developers of learning algorithms must inherently be aware of what the task goals are, yet we often require agents to discover them on their own without any supervision beyond these sparse rewards. W…
New method predicts and optimizes test-time scaling for LLMs.
problem Lack of principled guidance on scaling LLMs efficiently.
method Tail-guided search to predict and allocate compute.
result SLG Search achieves higher rewards with less compute.
New method for LLMs to learn reasoning by optimizing latent variables.
problem Teaching LLMs to generate logical justifications for answers.
method Formalized reasoning as latent variable model, derived FEM objective, designed sampling schemes.
result Prompt Posterior Sampling (PPS) outperforms other schemes in learning to reason.
Study designs logging policies to minimize off-policy evaluation error.
problem Minimizing OPE error with logging policies for target policies.
method Characterizes reward-coverage tradeoff, proposes a unifying framework, derives optimal policies.
result Provides actionable guidance for firms choosing recommendation systems.
This work formalizes guidance in diffusion models and introduces a stochastic control framework.
problem Lack of a solid theoretical foundation for guidance scheduling in diffusion models.
method Introduces a stochastic optimal control framework to cast guidance scheduling as an adaptive optimization problem.
result Establishes a principled foundation for more effective guidance in diffusion models.
A minimalist approach improves LLM reasoning by filtering incorrect responses.
problem Improving large language model (LLM) reasoning on complex tasks.
method Revisit GRPO from a reinforce-like algorithm perspective, proposing Reinforce-Rej.
result RAFT, a simple rejection sampling baseline, outperforms GRPO and PPO.
Framework for training-free guidance in discrete diffusion models for molecular generation.
problem No equivalent training-free guidance methods for discrete diffusion models.
method Framework using guidance functions for discrete data.
result Demonstrated utility on molecular graph generation tasks.
Study analyzes factors affecting profits in crypto liquidity provision.
problem Liquidity providers lack guidance for developing profitable strategies.
method Developed a measurement model based on impermanent loss to analyze key parameters.
result Uncovered influences of key parameters on LPs' profits.
Exact guidance for discrete data improves posterior sampling efficiency.
problem Inefficient guidance for discrete data in posterior sampling.
method Derive exact transition rate for desired distribution given learned discrete flow matching model.
result Significantly improved efficiency with single forward pass per sampling step.
Diffusion models improve sample quality with guidance, proving it works for general data distributions.
problem Theoretical understanding of guidance effect in diffusion models for general data distributions.
method Analyzing diffusion guidance under general data distributions, proving improvement in sample quality.
result Guidance improves the average reciprocal of the classifier probability, aligning with its motivation.
Research shows guidance in diffusion models does not sample from intended distribution, affecting boundary sampling.
problem Clarifying the misconception that guidance modifies the data distribution in diffusion models.
method Rigorous proof and fine-grained analysis of guidance dynamics in two cases: mixtures of compactly supported distributions and mixtures of Gaussians.
result Guidance leads to sampling more heavily from the boundary of the support of the conditional distribution as the parameter increases.
Regression problems assume every instance is annotated (labeled) with a real value, a form of annotation we call \emph{strong guidance}. In order for these annotations to be accurate, they must be the result of a precise experiment or measurement. However, in some cases additional \emph{weak guidance} might be given by…
Improved image quality in diffusion models by limiting guidance to a specific noise level range.
problem Improving image quality in diffusion models with guidance.
method Restricting guidance to a specific noise level range.
result Significantly improved FID score from 1.81 to 1.40 in ImageNet-512.
Unified approach to guided generation techniques.
problem Controlling the generative process of flow/diffusion models.
method Unified posterior and end-to-end guidance techniques.
result Unified posterior guidance as a greedy strategy of end-to-end guidance.
Enhances RL performance with a population-guided parallel learning scheme.
problem Improving off-policy reinforcement learning performance.
method Population-guided parallel learning scheme with shared experience replay buffer and soft policy update.
result Monotone improvement of the expected cumulative return proved theoretically and demonstrated in practice.
Jeffrey guidance extends diffusion-model control to more complex applications.
problem Controlling diffusion models beyond simple cases like conditional sampling.
method Leveraging Jeffrey's rule of conditioning to update marginal distributions towards a target distribution.
result Significant reductions in FID on CIFAR-10 and FFHQ with Inception embeddings as the target.
Reinforcement Learning (RL) is a widely employed machine learning architecture that has been applied to a variety of control problems. However, applications in safety-critical domains require a systematic and formal approach to specifying requirements as tasks or goals. We propose a model-free RL algorithm that enables…
The paper analyzes how guidance affects diffusion models using Gaussian mixture models.
problem Understanding how guidance influences diffusion models in specific contexts.
method Theoretical study using Gaussian mixture models and comparison inequalities for differential equations.
result Guidance boosts classification confidence but reduces distribution diversity, leading to lower differential entropy.
New study on guidance in masked diffusion models, showing how it shapes sampling dynamics.
problem Understanding how guidance influences the sampling behavior of masked diffusion models.
method Derived explicit solution to guided reverse dynamics, analyzing effects in 1D and 2D.
result Guidance amplifies class-specific regions and suppresses shared regions, affecting covariance structures.
The paper analyzes CFG in masked diffusion models and proposes a new method to improve sample quality.
problem Improving sample quality in conditional generation with CFG in masked diffusion models.
method Analyzing CFG's effect in a low-dimensional masked diffusion model, revealing early-stage guidance harms quality and late-stage improves it. Proposed a novel method to smooth transitions between distributions.
result The proposed method improves sample quality via a simple one-line code change.
SDG uses optimal control to improve classifier guidance in low-density regions.
problem Inefficient guidance in low-density regions of posterior distributions.
method Integrates stochastic optimal control with Stein variational inference to compute the steepest descent direction.
result SDG improves guidance in low-density regions, outperforming standard methods.
New method controls classifier guidance in diffusion models.
problem Improving classifier guidance in diffusion models.
method Cross-entropy control of classifier gradients.
result Effective guidance vectors with mean squared error O(dε). New method guides pretrained diffusion models without additional training.
problem Guidance methods for diffusion models often require extra training or are task-specific.
method Variational Control using Diffusion Trajectory Matching (DTM)
result Achieves state-of-the-art results on various problems.
Gradient guidance improves diffusion models for optimizing specific objectives.
problem Improving diffusion models for specific optimization tasks.
method Established a mathematical framework for gradient-guided diffusion, linking it to optimization theory. Developed a modified gradient guidance method and iteratively fine-tuned version.
result Gradient-guided diffusion models are essentially solutions to regularized optimization problems, preserving latent structure.
DTS improves inference-time alignment of diffusion models with less compute.
problem Inference-time alignment of diffusion models suffers from inaccurate value estimation and inefficient reuse of past computations.
method Diffusion Tree Sampling (DTS) uses a tree-based approach to propagate terminal rewards and iteratively refine value estimates.
result DTS produces asymptotically exact samples and matches the FID of best-performing baselines with up to 10x less compute.
Sharp statistical theory for conditional diffusion models.
problem Lack of theoretical foundation for conditional diffusion models.
method Sharp statistical theory with approximation of conditional score function.
result Sample complexity bound that adapts to data distribution smoothness.
The paper explores how to apply causal knowledge across different datasets to improve learning.
problem How to apply causal knowledge across different datasets to improve learning.
method Investigates the structural causal bandit with transportability, fusing priors from source environments to enhance learning in the deployment setting.
result Achieves a sub-linear regret bound with an explicit dependence on informativeness of prior data, potentially outperforming standard bandit approaches.
We present a reinforcement learning approach for detecting objects within an image. Our approach performs a step-wise deformation of a bounding box with the goal of tightly framing the object. It uses a hierarchical tree-like representation of predefined region candidates, which the agent can zoom in on. This reduces t…
Dynamic abstention improves LLM accuracy by selectively terminating unpromising reasoning.
problem LLMs waste compute on incorrect responses, leading to inefficiency.
method Formal reinforcement learning framework with abstention reward parameter.
result Dynamic abstention outperforms natural baselines in selective accuracy.