Active Learning, Multi-Objective Optimisation

e-PAL: An Active Learning Approach to the Multi-Objective Optimization Problem

Active Learning for Multi-Objective Optimizationm

The implementation code can be found here.

Active learning of Pareto fronts

Active learning of Pareto fronts with disconnected feasible

Synthesizing robust adversarial examples

Authors: Anish Athalye, Logan Engstrom, Andrew Ilyas, Kevin Kwok

Arxiv Link: https://arxiv.org/pdf/1707.07397.pdf

Algorithm parameter configuration

On continuous problems, it can select the optimal algorithm parameter. Per instance algorithm configuration of CMA-ES with limited budget and slides

A more completed method is presented on SAT, TSP and MIP problems. Algorithm runtime prediction: Methods & evaluation

Sequential Model-Based Optimization for General Algorithm Configuration This paper propose the SMAC algorithm, a random forest based Bayesian optimization algorithm.

Algorithm selection using deep learning without feature extraction

Reinforcement Learning

First of all, a list of resource for RL including courses, papers/surveys, benchmarks/testbeds Resources for Deep Reinforcement Learning

OpenAI tutorial Spinning Up in Deep RL

[Reinforcement Learning in Continuous State and Action Spaces]

Asynchronous Methods for Deep Reinforcement Learning

Addressing Function Approximation Error in Actor-Critic Methods

Proximal Policy Optimization Algorithms

Distributional Reinforcement Learning with Quantile Regression

Model-Based Value Expansion for Efficient Model-Free Reinforcement Learning

Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor

PPO-CMA: Proximal Policy Optimization with Covariance Matrix Adaptation

RL algorithms for handling combinatorial optimization problems Deep Learning in Computational Discrete Optimization

[Implicit Policy for Reinforcement Learning](

Improvements

Boosting Trust Region Policy Optimization with Normalizing Flows Policy

Stabilizing Off-Policy Reinforcement Learning with Conservative Policy Gradients

Hierarchical Reinforcement Learning

Data-Efficient Hierarchical Reinforcement Learning

Analysis

Visualizing Movement Control Optimization Landscapes

Black-box Optimization and Reinforcement Learning

Evolution Strategies as a Scalable Alternative to Reinforcement Learning blog

Guiding Evolutionary Strategies with Off-Policy Actor-Critic

Actor-critic versus direct policy search: a comparison based on sample complexity

Evolutionary Reinforcement Learning for Sample-Efficient Multiagent Coordination

Collaborative Evolutionary Reinforcement Learning source code.

Evolution-Guided Policy Gradient in Reinforcement Learning

Evolutionary Computation for Reinforcement Learning

CEM-RL: Combining evolutionary and gradient-based methods for policy search Reinforcement learning versus evolutionary computation: A survey on hybrid algorithms

Deep learning as Bayesian inference

[Dropout as a Bayesian Approximation]

High-Dimensional

[Dropout as a Bayesian Approximation]

Bandits

High-Dimensional Gaussian Process Bandits

Stochastic Multi-Armed-Bandit Problem with Non-stationary Rewards

Analysis of Thompson Sampling for the Multi-armed Bandit Problem

Thompson Sampling for Multi-Objective Multi-Armed Bandits Problem

Thompson Sampling for Contextual Bandits with Linear Payoffs

Natural Gradient

True Asymptotic Natural Gradient Optimization

Natural Langevin Dynamics for Neural Networks

Natural Gradient Deep Q-learning

New insights and perspectives on the natural gradient method First-order and second-order variants of the gradient descent: a unified framework

Natural Gradient for Black-box optimization

stochastic relaxition Natural Gradient, Fitness Modelling and Model Selection: A Unifying Perspective

Clustering

Efficient Parameter-free Clustering Using First Neighbor Relations