Defining and measuring decision-making styles, also known as playstyles, is crucial in gaming, where these styles reflect a broad spectrum of individuality and diversity. However, finding a universally applicable measure for these styles poses a challenge. Building on Playstyle Distance, the first unsupervised metric to measure playstyle similarity based on game screens and raw actions, we introduce three enhancements to increase accuracy: multiscale analysis with varied state granularity, a perceptual kernel rooted in psychology, and the utilization of the intersection-over-union method for efficient evaluation. These innovations not only advance measurement precision but also offer insights into human cognition of similarity. Across two racing games and seven Atari games, our techniques significantly improve the precision of zero-shot playstyle classification, achieving an accuracy exceeding 90 percent with fewer than 512 observation-action pairs, which is less than half an episode of these games. Furthermore, our experiments with 2048 and Go demonstrate the potential of discrete playstyle measures in puzzle and board games. We also develop an algorithm for assessing decision-making diversity using these measures. Our findings improve the measurement of end-to-end game analysis and the evolution of artificial intelligence for diverse playstyles.
Monte-Carlo Tree Search (MCTS) methods, such as Upper Confidence Bound applied to Trees (UCT), are instrumental to automated planning techniques. However, UCT can be slow to explore an optimal action when it initially appears inferior to other actions. Maximum ENtropy Tree-Search (MENTS) incorporates the maximum entropy principle into an MCTS approach, utilising Boltzmann policies to sample actions, naturally encouraging more exploration. In this paper, we highlight a major limitation of MENTS: optimal actions for the maximum entropy objective do not necessarily correspond to optimal actions for the original objective. We introduce two algorithms, Boltzmann Tree Search (BTS) and Decaying ENtropy Tree-Search (DENTS), that address these limitations and preserve the benefits of Boltzmann policies, such as allowing actions to be sampled faster by using the Alias method. Our empirical analysis shows that our algorithms show consistent high performance across several benchmark domains, including the game of Go.
Reinforcement learning (RL) is a powerful tool for optimal control that has found great success in Atari games, the game of Go, robotic control, and building optimization. RL is also very brittle; agents often overfit to their training environment and fail to generalize to new settings. Unsupervised environment design (UED) has been proposed as a solution to this problem, in which the agent trains in environments that have been specially selected to help it learn. Previous UED algorithms focus on trying to train an RL agent that generalizes across a large distribution of environments. This is not necessarily desirable when we wish to prioritize performance in one environment over others. In this work, we will be examining the setting of robust RL building control, where we wish to train an RL agent that prioritizes performing well in normal weather while still being robust to extreme weather conditions. We demonstrate a novel UED algorithm, ActivePLR, that uses uncertainty-aware neural network architectures to generate new training environments at the limit of the RL agent's ability while being able to prioritize performance in a desired base environment. We show that ActivePLR is able to outperform state-of-the-art UED algorithms in minimizing energy usage while maximizing occupant comfort in the setting of building control.
The success of AlphaZero (AZ) has demonstrated that neural-network-based Go AIs can surpass human performance by a large margin. Given that the state space of Go is extremely large and a human player can play the game from any legal state, we ask whether adversarial states exist for Go AIs that may lead them to play surprisingly wrong actions. In this paper, we first extend the concept of adversarial examples to the game of Go: we generate perturbed states that are ``semantically'' equivalent to the original state by adding meaningless moves to the game, and an adversarial state is a perturbed state leading to an undoubtedly inferior action that is obvious even for Go beginners. However, searching the adversarial state is challenging due to the large, discrete, and non-differentiable search space. To tackle this challenge, we develop the first adversarial attack on Go AIs that can efficiently search for adversarial states by strategically reducing the search space. This method can also be extended to other board games such as NoGo. Experimentally, we show that the actions taken by both Policy-Value neural network (PV-NN) and Monte Carlo tree search (MCTS) can be misled by adding one or two meaningless stones; for example, on 58\% of the AlphaGo Zero self-play games, our method can make the widely used KataGo agent with 50 simulations of MCTS plays a losing action by adding two meaningless stones. We additionally evaluated the adversarial examples found by our algorithm with amateur human Go players and 90\% of examples indeed lead the Go agent to play an obviously inferior action. Our code is available at \url{https://PaperCode.cc/GoAttack}.
Reinforcement Learning (RL) has been able to solve hard problems such as playing Atari games or solving the game of Go, with a unified approach. Yet modern deep RL approaches are still not widely used in real-world applications. One reason could be the lack of guarantees on the performance of the intermediate executed policies, compared to an existing (already working) baseline policy. In this paper, we propose an online model-free algorithm that solves conservative exploration in the policy optimization problem. We show that the regret of the proposed approach is bounded by $\tilde{\mathcal{O}}(\sqrt{T})$ for both discrete and continuous parameter spaces.
MuZero, a model-based reinforcement learning algorithm that uses a value equivalent dynamics model, achieved state-of-the-art performance in Chess, Shogi and the game of Go. In contrast to standard forward dynamics models that predict a full next state, value equivalent models are trained to predict a future value, thereby emphasizing value relevant information in the representations. While value equivalent models have shown strong empirical success, there is no research yet that visualizes and investigates what types of representations these models actually learn. Therefore, in this paper we visualize the latent representation of MuZero agents. We find that action trajectories may diverge between observation embeddings and internal state transition dynamics, which could lead to instability during planning. Based on this insight, we propose two regularization techniques to stabilize MuZero's performance. Additionally, we provide an open-source implementation of MuZero along with an interactive visualizer of learned representations, which may aid further investigation of value equivalent algorithms.
The widespread availability of superhuman AI engines is changing how we play the ancient game of Go. The open-source software packages developed after the AlphaGo series shifted focus from producing strong playing entities to providing tools for analyzing games. Here we describe two ways of how the innovations of the second generation engines (e.g.~score estimates, variable komi) can be used for defining new metrics that help deepen our understanding of the game. First, we study how much information the search component contributes in addition to the raw neural network policy output. This gives an intrinsic strength measurement for the neural network. Second, we define the effect of a move by the difference in score estimates. This gives a fine-grained, move-by-move performance evaluation of a player. We use this in combating the new challenge of detecting online cheating.
Deep learning's recent history has been one of achievement: from triumphing over humans in the game of Go to world-leading performance in image classification, voice recognition, translation, and other tasks. But this progress has come with a voracious appetite for computing power. This article catalogs the extent of this dependency, showing that progress across a wide variety of applications is strongly reliant on increases in computing power. Extrapolating forward this reliance reveals that progress along current lines is rapidly becoming economically, technically, and environmentally unsustainable. Thus, continued progress in these applications will require dramatically more computationally-efficient methods, which will either have to come from changes to deep learning or from moving to other machine learning methods.
Rapid advancements in deep learning have led to many recent breakthroughs. While deep learning models achieve superior performance, often statistically better than humans, their adoption into safety-critical settings, such as healthcare or self-driving cars is hindered by their inability to provide safety guarantees or to expose the inner workings of the model in a human understandable form. We present Mo\"ET, a novel model based on Mixture of Experts, consisting of decision tree experts and a generalized linear model gating function. Thanks to such gating function the model is more expressive than the standard decision tree. To support non-differentiable decision trees as experts, we formulate a novel training procedure. In addition, we introduce a hard thresholding version, Mo\"ETH, in which predictions are made solely by a single expert chosen via the gating function. Thanks to that property, Mo\"ETH allows each prediction to be easily decomposed into a set of logical rules in a form which can be easily verified. While Mo\"ET is a general use model, we illustrate its power in the reinforcement learning setting. By training Mo\"ET models using an imitation learning procedure on deep RL agents we outperform the previous state-of-the-art technique based on decision trees while preserving the verifiability of the models. Moreover, we show that Mo\"ET can also be used in real-world supervised problems on which it outperforms other verifiable machine learning models.
By introducing several improvements to the AlphaZero process and architecture, we greatly accelerate self-play learning in Go, achieving a 50x reduction in computation over comparable methods. Like AlphaZero and replications such as ELF OpenGo and Leela Zero, our bot KataGo only learns from neural-net-guided Monte Carlo tree search self-play. But whereas AlphaZero required thousands of TPUs over several days and ELF required thousands of GPUs over two weeks, KataGo surpasses ELF's final model after only 19 days on fewer than 30 GPUs. Much of the speedup involves non-domain-specific improvements that might directly transfer to other problems. Further gains from domain-specific techniques reveal the remaining efficiency gap between the best methods and purely general methods such as AlphaZero. Our work is a step towards making learning in state spaces as large as Go possible without large-scale computational resources.
The AlphaGo, AlphaGo Zero, and AlphaZero series of algorithms are remarkable demonstrations of deep reinforcement learning's capabilities, achieving superhuman performance in the complex game of Go with progressively increasing autonomy. However, many obstacles remain in the understanding of and usability of these promising approaches by the research community. Toward elucidating unresolved mysteries and facilitating future research, we propose ELF OpenGo, an open-source reimplementation of the AlphaZero algorithm. ELF OpenGo is the first open-source Go AI to convincingly demonstrate superhuman performance with a perfect (20:0) record against global top professionals. We apply ELF OpenGo to conduct extensive ablation studies, and to identify and analyze numerous interesting phenomena in both the model training and in the gameplay inference procedures. Our code, models, selfplay datasets, and auxiliary data are publicly available at https://ai.facebook.com/tools/elf-opengo/.
Competing with top human players in the ancient game of Go has been a long-term goal of artificial intelligence. Go's high branching factor makes traditional search techniques ineffective, even on leading-edge hardware, and Go's evaluation function could change drastically with one stone change. Recent works [Maddison et al. (2015); Clark & Storkey (2015)] show that search is not strictly necessary for machine Go players. A pure pattern-matching approach, based on a Deep Convolutional Neural Network (DCNN) that predicts the next move, can perform as well as Monte Carlo Tree Search (MCTS)-based open source Go engines such as Pachi [Baudis & Gailly (2012)] if its search budget is limited. We extend this idea in our bot named darkforest, which relies on a DCNN designed for long-term predictions. Darkforest substantially improves the win rate for pattern-matching approaches against MCTS-based approaches, even with looser search budgets. Against human players, the newest versions, darkfores2, achieve a stable 3d level on KGS Go Server as a ranked bot, a substantial improvement upon the estimated 4k-5k ranks for DCNN reported in Clark & Storkey (2015) based on games against other machine players. Adding MCTS to darkfores2 creates a much stronger player named darkfmcts3: with 5000 rollouts, it beats Pachi with 10k rollouts in all 250 games; with 75k rollouts it achieves a stable 5d level in KGS server, on par with state-of-the-art Go AIs (e.g., Zen, DolBaram, CrazyStone) except for AlphaGo [Silver et al. (2016)]; with 110k rollouts, it won the 3rd place in January KGS Go Tournament.