Abstract:Machine learning processes, e.g. ''learning in games'', can be viewed as non-linear dynamical systems. In general, such systems exhibit a wide spectrum of behaviors, ranging from stability/recurrence to the undesirable phenomena of chaos (or ''butterfly effect''). Chaos captures sensitivity of round-off errors and can severely affect predictability and reproducibility of ML systems, but AI/ML community's understanding of it remains rudimentary. It has a lot out there that await exploration. Recently, Cheung and Piliouras employed volume-expansion argument to show that Lyapunov chaos occurs in the cumulative payoff space, when some popular learning algorithms, including Multiplicative Weights Update (MWU), Follow-the-Regularized-Leader (FTRL) and Optimistic MWU (OMWU), are used in several subspaces of games, e.g. zero-sum, coordination or graphical constant-sum games. It is natural to ask: can these results generalize to much broader families of games? We take on a game decomposition approach and answer the question affirmatively. Among other results, we propose a notion of ''matrix domination'' and design a linear program, and use them to characterize bimatrix games where MWU is Lyapunov chaotic almost everywhere. Such family of games has positive Lebesgue measure in the bimatrix game space, indicating that chaos is a substantial issue of learning in games. For multi-player games, we present a local equivalence of volume change between general games and graphical games, which is used to perform volume and chaos analyses of MWU and OMWU in potential games.

Programs with Stringent Performance Objectives Will Often Exhibit Chaotic Behavior

Chess Pure Strategies are Probably Chaotic

Chaos in computer performance

On game psychology: an experiment on the chess board/screen, should you always "do your best", and why the programs with prescribed weaknesses cannot be our good friends?

Chaos controlling of a kind of mechanical vibration systems

Chaos of Learning Beyond Zero-sum and Coordination via Game Decompositions

"Illusion of control" in Minority and Parrondo Games

Performance Evaluation

Chaotic Genetic Algorithm and The Effects of Entropy in Performance Optimization

Cooperative and uncooperative institution designs: Surprises and problems in open-source game theory

Analysis of Decision-Making in Economic Chaos Control

Memory loss can prevent chaos in games dynamics

Chaotic Monte Carlo Computation: A Dynamical Effect of Random-Number Generations

The Similarity of Dynamic behavior between Different Chaotic Systems

You Can't Always Get What You Want : Games of Ordered Preference

The Elliptical Envelope

Chaotic Binarization Schemes for Solving Combinatorial Optimization Problems Using Continuous Metaheuristics

Comparative analysis of two novel chaotic systems, and validation of hybrid function projective and complete synchronization using active-adaptive control

Chaos Control And Synchronization By Using An Intermittent Feedback

Delayed Feedback Chaotification Model and Its Hardware Implementation

CONTROLLING THE ULTIMATE STATE OF PROJECTIVE SYNCHRONIZATION IN CHAOS: APPLICATION TO CHAOTIC ENCRYPTION