Simple policy outperforms complex ones in cloud auto-scaling.
problem Predicting resource scaling for large-scale cloud applications with limited deployment throughput.
method Probabilistic workload forecast for auto-scaling decisions based on risk aversion.
result The proposed policy outperforms sophisticated and simple benchmark policies in real-world and synthetic data.
Simple policy search outperforms advanced learnable test-time augmentation techniques.
problem Improving predictive performance through test-time data augmentation.
method Greedy policy search (GPS) for learning test-time augmentation policies.
result Augmentation policies learned with GPS achieve superior predictive performance and robustness.
Optimistic NPG improves policy optimization in online RL with efficient sample complexity.
problem Limited theoretical understanding of policy optimization, especially in online RL.
method Combines natural policy gradient with optimistic policy evaluation.
result Achieves optimal dimension dependence sample complexity for learning near-optimal policies.
A new imitation learning method uses random search for simple policies, outperforming complex models.
problem Complexity and reward dependency issues in imitation learning.
method Derivative-free optimization with simple linear policies and random search.
result The proposed method achieves competitive performance on MuJoCo locomotion tasks without a direct reward signal.
AWR simplifies RL with simple supervised learning.
problem Developing scalable RL algorithms for off-policy data.
method Two supervised learning steps: value function and policy.
result AWR achieves competitive performance and effective policies.
On-policy reinforcement learning (RL) algorithms have high sample complexity while off-policy algorithms are difficult to tune. Merging the two holds the promise to develop efficient algorithms that generalize across diverse environments. It is however challenging in practice to find suitable hyper-parameters that gove…
Study on variance of policy gradient in simple RL environments.
problem Understanding variance of policy gradient estimators in continuous RL.
method Analyzes REINFORCE estimator in linear-quadratic environments with Gaussian noise.
result Derives and validates bounds on estimator variance empirically.
Paper proves exponential lower bounds for policy iteration in MDPs.
problem Proving lower bounds for policy iteration in multi-action MDPs.
method Generalized previous results to k-action MDPs, constructed families of MDPs.
result Proved novel exponential lower bound of (3+k)2^(N/2-3) iterations.
FOCOPS optimizes agent's behavior while adhering to constraints.
problem Optimizing agent's behavior while respecting safety constraints.
method FOCOPS solves a constrained optimization problem in policy space, then projects the solution back into the parametric space.
result FOCOPS achieves better performance on constrained robotics tasks.
In this work, we consider the problem of estimating a behaviour policy for use in Off-Policy Policy Evaluation (OPE) when the true behaviour policy is unknown. Via a series of empirical studies, we demonstrate how accurate OPE is strongly dependent on the calibration of estimated behaviour policy models: how precisely …
DNN policies improve stochastic AC OPF for power grid optimization.
problem Optimizing power grid operations under uncertainty.
method Deep neural network (DNN) policies for real-time generator dispatch decisions.
result DNN policies enforce feasibility constraints and produce near optimal solutions.
A streamlined DRL algorithm improves sample efficiency without entropy maximization.
problem Improving sample efficiency in off-policy DRL algorithms.
method Output normalization and non-uniform sampling.
result Proposed algorithm matches SAC's performance without entropy maximization and improves sample efficiency.
A classic setting of the stochastic K-armed bandit problem is considered in this note. In this problem it has been known that KL-UCB policy achieves the asymptotically optimal regret bound and KL-UCB+ policy empirically performs better than the KL-UCB policy although the regret bound for the original form of the KL-UCB…
SCPO learns robust policies without modeling disturbance, improving real-world task performance.
problem Poor performance of reinforcement learning in real-world tasks due to disturbance in transition dynamics.
method State-conservative policy optimization (SCPO) that reduces disturbance to state space and approximates it with a gradient-based regularizer.
result SCPO learns robust policies without prior knowledge of disturbance or simulators, improving performance in robot control tasks.
A single policy suffices for near-optimal parallel exploration in RL.
problem Quantitative effects of parallel exploration in reward-free RL.
method Using a single policy to guide exploration across all agents.
result Near-linear speedup and near-minimax optimality for linear MDPs.
Policy gradient methods achieve linear convergence in simple MDPs.
problem Analyzing convergence rates of policy gradient methods in finite MDPs.
method Connections with policy iteration to show linear convergence with large step-sizes.
result Policy gradient methods succeed with large step-sizes and achieve linear rate of convergence.
We propose an estimator and confidence interval for computing the value of a policy from off-policy data in the contextual bandit setting. To this end we apply empirical likelihood techniques to formulate our estimator and confidence interval as simple convex optimization problems. Using the lower bound of our confiden…
Study uses contextual bandits to optimize charity exposure in donation solicitation.
problem Optimizing charity exposure in donation solicitation using survey responses.
method Adaptive experiment design to balance cumulative regret minimization and simple regret minimization.
result Adaptive experimentation yields better policy learning outcomes than uniform randomization.
AI models outperform simple rules in cross-asset futures timing, especially with lower transaction costs.
problem Optimizing cross-asset portfolio weights using traditional forecasting and optimization methods.
method End-to-end AI policies that map market states directly to portfolio weights, trained on CME futures using a differentiable Sharpe ratio loss function.
result Transformer-based AI policies outperform simple rules and equal weighting, trading less and matching or exceeding equal weighting through moderate transaction costs.
Paper tackles offline SSP with value iteration for policy evaluation and learning.
problem Goal-oriented RL with offline data and cost minimization.
method Simple value iteration algorithms for OPE and offline policy learning.
result Strong instance-dependent bounds implying near-minimax optimal worst-case bounds.
SUNRISE improves off-policy RL algorithms by integrating ensemble methods.
problem Stability and exploration issues in off-policy RL algorithms.
method SUNRISE combines ensemble-based weighted Bellman backups and upper-confidence bounds for efficient exploration.
result SUNRISE improves the performance of off-policy RL algorithms across various domains.
We present the first differentially private algorithms for reinforcement learning, which apply to the task of evaluating a fixed policy. We establish two approaches for achieving differential privacy, provide a theoretical analysis of the privacy and utility of the two algorithms, and show promising results on simple e…
This work adopts the very successful distributional perspective on reinforcement learning and adapts it to the continuous control setting. We combine this within a distributed framework for off-policy learning in order to develop what we call the Distributed Distributional Deep Deterministic Policy Gradient algorithm, …
We observe that several existing policy gradient methods (such as vanilla policy gradient, PPO, A2C) may suffer from overly large gradients when the current policy is close to deterministic (even in some very simple environments), leading to an unstable training process. To address this issue, we propose a new method, …
Policy gradient methods are widely used for control in reinforcement learning, particularly for the continuous action setting. There have been a host of theoretically sound algorithms proposed for the on-policy setting, due to the existence of the policy gradient theorem which provides a simplified form for the gradien…
Improves RL algorithms with two techniques.
problem Enhance off-policy RL performance.
method Formulates RL as proximal point iteration; uses value functions for improved action value estimate.
result Significant performance improvement on RL benchmarks.
This paper investigates a type of instability that is linked to the greedy policy improvement in approximated reinforcement learning. We show empirically that non-deterministic policy improvement can stabilize methods like LSPI by controlling the improvements' stochasticity. Additionally we show that a suitable represe…
Simplifies RL training with fewer techniques, reducing bias and instability.
problem Training instabilities and high sample complexity in RL.
method Introduced a simple deterministic policy gradient, used propensity estimation, and delayed policy updates.
result Improved performance and reduced sample complexity through these techniques.
New algorithm for sequential off-policy learning improves performance over batch methods.
problem Training policies from logged interaction data in a sequential setting.
method Combines Logarithmic Smoothing with online PAC-Bayesian tools.
result Improves performance and accelerates convergence in sequential off-policy learning.
Algorithm learns to solve tasks by gradually expanding policy space.
problem Reinforcement learning difficulty progression.
method Decreasingly constrained policy space expansion.
result Superior learning rate in Tetris tasks.
ES-MAML uses Evolution Strategies for MAML, avoiding second derivative estimation.
problem Solving the MAML problem with efficient second derivative estimation.
method Applies Evolution Strategies to MAML, avoiding second derivative estimation.
result ES-MAML performs competitively and often better with fewer queries.
A generative recurrent neural network is quickly trained in an unsupervised manner to model popular reinforcement learning environments through compressed spatio-temporal representations. The world model's extracted features are fed into compact and simple policies trained by evolution, achieving state of the art resul…
Deep Reinforcement Learning (DRL) algorithms for continuous action spaces are known to be brittle toward hyperparameters as well as \cut{being}sample inefficient. Soft Actor Critic (SAC) proposes an off-policy deep actor critic algorithm within the maximum entropy RL framework which offers greater stability and empiric…
An important problem in sequential decision-making under uncertainty is to use limited data to compute a safe policy, i.e., a policy that is guaranteed to perform at least as well as a given baseline strategy. In this paper, we develop and analyze a new model-based approach to compute a safe policy when we have access …
New PG method derived from IS, improving variance reduction.
problem Improving policy gradient methods for better performance.
method Starting from the DR estimator, a new PG method is derived.
result Empirically shows the new DR-PG estimator is effective.
Study proposes a new approval policy for ML-based medical devices to prevent gradual performance degradation.
problem Gradual deterioration in machine learning model performance over time in medical devices.
method Formulated an automatic algorithmic change protocol (aACP) as an online hypothesis testing problem, considering both error-rate guarantees and non-guaranteed policies.
result Controlled the rate of gradual deterioration (biocreep) in machine learning models without significantly impacting approval of beneficial modifications.
Unified framework for policy learning using weak supervision.
problem High-quality supervision is often infeasible or expensive in practice.
method Treat weak supervision as imperfect peer information and evaluate policies based on correlated agreement.
result Substantial performance improvements, especially in complex or noisy environments.
Deep reinforcement learning has led to several recent breakthroughs, though the learned policies are often based on black-box neural networks. This makes them difficult to interpret and to impose desired specification constraints during learning. We present an iterative framework, MORL, for improving the learned polici…
End-to-end policy learning method improves CATE estimation.
problem Learning optimal treatment policies from partially observed data.
method Modified causal forest for policy learning.
result Maximizing policy value is equivalent to minimizing CATE.
Proposes fair and robust methods for estimating treatment effects.
problem Estimating treatment effects while maintaining fairness.
method Simple, nonparametric framework with fairness constraints.
result Estimators are double robust and characterize welfare trade-offs.
While recent progress has spawned very powerful machine learning systems, those agents remain extremely specialized and fail to transfer the knowledge they gain to similar yet unseen tasks. In this paper, we study a simple reinforcement learning problem and focus on learning policies that encode the proper invariances …
Combines BC and GAIL for efficient imitation learning.
problem Efficient imitation learning without reward signals.
method Integrates Behavior Cloning and Generative Adversarial Imitation Learning.
result Combination leads to stable and sample-efficient learning.
POLAR learns efficient data acquisition policies using pretrained belief representations.
problem Challenges in learning effective policies for adaptive data acquisition.
method POLAR decouples representation learning from policy learning by leveraging pretrained predictive foundation models as belief-state encoders.
result POLAR outperforms state-of-the-art methods across diverse tasks while requiring fewer training samples.
Learning goal-oriented dialogues by means of deep reinforcement learning has recently become a popular research topic. However, commonly used policy-based dialogue agents often end up focusing on simple utterances and suboptimal policies. To mitigate this problem, we propose a class of novel temperature-based extension…
Policy gradients methods apply to complex, poorly understood, control problems by performing stochastic gradient descent over a parameterized class of polices. Unfortunately, even for simple control problems solvable by standard dynamic programming techniques, policy gradient algorithms face non-convex optimization pro…
Simplifies complex RL policies by ranking important decisions.
problem Complexity in RL policies makes them hard to analyze and interpret.
method Statistical fault localisation to rank states and prune unimportant decisions.
result Pruned policies can perform similarly to original policies, improving interpretability.
The study compares reinforcement learning models and finds model-based approaches superior for complex MDPs.
problem Complexity of optimal Q-functions and policies in MDPs exceeds dynamics, hindering model-free methods.
method Theoretical analysis and empirical testing of neural network expressivity for policies, Q-functions, and dynamics.
result Model-based planning yields better policies for complex MDPs, improving performance on MuJoCo tasks.
AI-Interpret transforms opaque policies into simple, interpretable decision rules.
problem Designing effective decision aids for professionals to mitigate decision-making biases.
method Combining imitation learning, program induction, and clustering to transform learned policies into interpretable descriptions.
result Providing interpretable decision rules as flowcharts significantly improves people's planning strategies and decisions.