arXiv research

Paper applies NFSP to Mini-RTS, a small RTS game.

Paper improves autonomous vehicle safety and efficiency with new reinforcement learning methods.

Expanding models in neural fictitious play improves reinforcement learning efficiency and robustness.

Generative adversarial networks (GANs) are powerful tools for learning generative models. In practice, the training may suffer from lack of convergence. GANs are commonly viewed as a two-player zero-sum game between two neural networks. Here, we leverage this game theoretic view to study the convergence behavior of the…

ESCHER avoids importance sampling to estimate regret in large games.

Deep learning theory for Nash equilibrium in stochastic games.

KataGo accelerates Go self-play learning by 50x.

Paper addresses theoretical risks in neural MCCFR, proposing Robust Deep MCCFR for improved performance.

Fictitious play is a simple and widely studied adaptive heuristic for playing repeated games. It is well known that fictitious play fails to be Hannan consistent. Several variants of fictitious play including regret matching, generalized regret matching and smooth fictitious play, are known to be Hannan consistent. In …

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…

New algorithm improves self-play reinforcement learning for competitive games.

We present a simple model of a non-equilibrium self-organizing market where asset prices are partially driven by investment decisions of a bounded-rational agent. The agent acts in a stochastic market environment driven by various exogenous "alpha" signals, agent's own actions (via market impact), and noise. Unlike tra…

Adversarial self-play in two-player games has delivered impressive results when used with reinforcement learning algorithms that combine deep neural networks and tree search. Algorithms like AlphaZero and Expert Iteration learn tabula-rasa, producing highly informative training data on the fly. However, the self-play t…

Deep neural network solves large multi-agent games for Markovian Nash equilibrium.

It is now well known that decentralised optimisation can be formulated as a potential game, and game-theoretical learning algorithms can be used to find an optimum. One of the most common learning techniques in game theory is fictitious play. However fictitious play is founded on an implicit assumption that opponents' …

Automatically finds effective security strategies through reinforcement learning and self-play.

New algorithms prove self-play can be effective in competitive RL.

Deep fictitious play converges to Nash equilibrium in stochastic differential games.

This paper improves self-play learning in games by manipulating experience distributions.

Novel analysis of EFP for finite-sum problems in neural networks.

Decentralised optimisation tasks are important components of multi-agent systems. These tasks can be interpreted as n-player potential games: therefore game-theoretic learning algorithms can be used to solve decentralised optimisation tasks. Fictitious play is the canonical example of these algorithms. Nevertheless fic…

Framework improves self-play for cooperative multi-agent learning.

New method labels GAN-generated faces without stereotyping.

The paper investigates how supervised learning and self-play improve sample efficiency in teaching AI to communicate.

The paper uses linear function approximators to bias MCTS for general games.

AlphaZero's reward function is replaced with a total ordering, enabling self-play without balancing.

New algorithm reduces MFGs with common noise complexity.

We introduce a new virtual environment for simulating a card game known as "Big 2". This is a four-player game of imperfect information with a relatively complicated action space (being allowed to play 1,2,3,4 or 5 card combinations from an initial starting hand of 13 cards). As such it poses a challenge for many curre…

Bayesian framework improves trading robustness against market shifts.

SPIN converts weak LLMs to strong ones using self-play.

The paper develops algorithms for competitive RL in partially observable MGs.

Many poker systems, whether created with heuristics or machine learning, rely on the probability of winning as a key input. However calculating the precise probability using combinatorics is an intractable problem, so instead we approximate it. Monte Carlo simulation is an effective technique that can be used to approx…

New method learns optimal environment and goal difficulty for reinforcement learning.

Two new RL methods enable deep learning of MFG equilibria.

Deep reinforcement learning optimizes retrosynthetic planning for chemical synthesis.

The paper simplifies complex jump-diffusion markets to complete models.

Self-play fine-tuning improves diffusion models for text-to-image generation.

Paper optimizes reinforcement learning in self-play games with reduced steps.

Paper proposes a self-play method to approximate human evaluation of conversational agents.

ISP improves DNN uncertainty for better subgroup accuracy.

Study learns optimal strategies in imperfect information games with self-play.

The paper proposes a new approach to model risk measurement based on the Wasserstein distance between two probability measures. It formulates the theoretical motivation resulting from the interpretation of fictitious adversary of robust risk management. The proposed approach accounts for equivalent and non-equivalent p…

In hierarchical reinforcement learning a major challenge is determining appropriate low-level policies. We propose an unsupervised learning scheme, based on asymmetric self-play from Sukhbaatar et al. (2018), that automatically learns a good representation of sub-goals in the environment and a low-level policy that can…

The paper proposes a new policy training objective to reduce exploration in self-play.

SPPO optimizes language model alignment by treating preferences as a game and achieving state-of-the-art performance.

Develops a new theory for neural systems stability and width effects.

This paper analyzes convergence of model-free learning in Mean Field Games.

We consider a family of learning strategies for online optimization problems that evolve in continuous time and we show that they lead to no regret. From a more traditional, discrete-time viewpoint, this continuous-time approach allows us to derive the no-regret properties of a large class of discrete-time algorithms i…