arXiv subjects

Viliam Lisy

Publications and source records attributed to Viliam Lisy.

12 recordsLinked to original sources

arXiv2026

Test-time Reinforcement Learning in Imperfect Information Games

Test-time reasoning has significantly improved performance in domains ranging from games to language models. However, test-time policy changes with formal guarantees on the performance of the resulting strategy remain a challenge in two-player zero-sum imperfect-information games. Existing solutions are limited to tabular methods or single gradient step updates. In this work, we investigate policy-gradient algorithms as a method for scalable test-time reasoning. We extend the concept of gadget game, tabular technique for test-time search, to the reinforcement learning setting. Unlike prior approaches, we represent the gadget game implicitly by modified sampling and neural policy rather then explicitly by constructing it, thereby removing constraints on subgame size. Furthermore, we formally prove that, unlike prior tabular algorithms, regularized policy-gradient algorithms limit possible strategy degradation caused by test-time reasoning, even without the gadget games. Our evaluation across small- and large-scale games confirms that additional test-time training often substantially improves performance relative to the blueprint strategy.

cs.GT↗

arXiv2026

Equilibrium Refinements Improve Subgame Solving in Imperfect-Information Games

Subgame solving is a technique for scaling algorithms to large games by locally refining a precomputed blueprint strategy during gameplay. While straightforward in perfect-information games where search starts from the current state, subgame solving in imperfect-information games must account for hidden states and uncertainty about the opponent's past strategy. Gadget games were developed to ensure that the improved subgame strategy is robust against any possible opponent's strategy in a zero-sum game. Gadget games typically contain infinitely many Nash equilibria. We demonstrate that while these equilibria are equivalent in the gadget game, they yield vastly different performance in the full game, even when facing a rational opponent. We propose gadget game sequential equilibria as the preferred solution concept. We introduce modifications to the sequence-form linear program and counterfactual regret minimization that converge to these refined solutions with only mild additional computational cost. Additionally, we provide several new insights into the surprising superiority of the resolving gadget game over the max-margin gadget game. Our experiments compare different Nash equilibria of gadget games in several standard benchmark games, showing that our refined equilibria consistently outperform unrefined Nash equilibria, and can reduce the exploitability of the overall strategy by more than 50%

cs.GT↗

arXiv2024

Counteracting Concept Drift by Learning with Future Malware Predictions

The accuracy of deployed malware-detection classifiers degrades over time due to changes in data distributions and increasing discrepancies between training and testing data. This phenomenon is known as the concept drift. While the concept drift can be caused by various reasons in general, new malicious files are created by malware authors with a clear intention of avoiding detection. The existence of the intention opens a possibility for predicting such future samples. Including predicted samples in training data should consequently increase the accuracy of the classifiers on new testing data. We compare two methods for predicting future samples: (1) adversarial training and (2) generative adversarial networks (GANs). The first method explicitly seeks for adversarial examples against the classifier that are then used as a part of training data. Similarly, GANs also generate synthetic training data. We use GANs to learn changes in data distributions within different time periods of training data and then apply these changes to generate samples that could be in testing data. We compare these prediction methods on two different datasets: (1) Ember public dataset and (2) the internal dataset of files incoming to Avast. We show that while adversarial training yields more robust classifiers, this method is not a good predictor of future malware in general. This is in contrast with previously reported positive results in different domains (including natural language processing and spam detection). On the other hand, we show that GANs can be successfully used as predictors of future malware. We specifically examine malware families that exhibit significant changes in their data distributions over time and the experimental results confirm that GAN-based predictions can significantly improve the accuracy of the classifier on new, previously unseen data.

cs.CR↗

arXiv2023

Look-ahead Search on Top of Policy Networks in Imperfect Information Games

Search in test time is often used to improve the performance of reinforcement learning algorithms. Performing theoretically sound search in fully adversarial two-player games with imperfect information is notoriously difficult and requires a complicated training process. We present a method for adding test-time search to an arbitrary policy-gradient algorithm that learns from sampled trajectories. Besides the policy network, the algorithm trains an additional critic network, which estimates the expected values of players following various transformations of the policies given by the policy network. These values are then used for depth-limited search. We show how the values from this critic can create a value function for imperfect information games. Moreover, they can be used to compute the summary statistics necessary to start the search from an arbitrary decision point in the game. The presented algorithm is scalable to very large games since it does not require any search during train time. We evaluate the algorithm's performance when trained along Regularized Nash Dynamics, and we evaluate the benefit of using the search in the standard benchmark game of Leduc hold'em, multiple variants of imperfect information Goofspiel, and Battleships.

cs.GT↗

arXiv2022

Avast-CTU Public CAPE Dataset

There is a limited amount of publicly available data to support research in malware analysis technology. Particularly, there are virtually no publicly available datasets generated from rich sandboxes such as Cuckoo/CAPE. The benefit of using dynamic sandboxes is the realistic simulation of file execution in the target machine and obtaining a log of such execution. The machine can be infected by malware hence there is a good chance of capturing the malicious behavior in the execution logs, thus allowing researchers to study such behavior in detail. Although the subsequent analysis of log information is extensively covered in industrial cybersecurity backends, to our knowledge there has been only limited effort invested in academia to advance such log analysis capabilities using cutting edge techniques. We make this sample dataset available to support designing new machine learning methods for malware detection, especially for automatic detection of generic malicious behavior. The dataset has been collected in cooperation between Avast Software and Czech Technical University - AI Center (AIC).

cs.CR↗

arXiv2021

Improving Robustness of Malware Classifiers using Adversarial Strings Generated from Perturbed Latent Representations

In malware behavioral analysis, the list of accessed and created files very often indicates whether the examined file is malicious or benign. However, malware authors are trying to avoid detection by generating random filenames and/or modifying used filenames with new versions of the malware. These changes represent real-world adversarial examples. The goal of this work is to generate realistic adversarial examples and improve the classifier's robustness against these attacks. Our approach learns latent representations of input strings in an unsupervised fashion and uses gradient-based adversarial attack methods in the latent domain to generate adversarial examples in the input domain. We use these examples to improve the classifier's robustness by training on the generated adversarial set of strings. Compared to classifiers trained only on perturbed latent vectors, our approach produces classifiers that are significantly more robust without a large trade-off in standard accuracy.

cs.LG↗

arXiv2021

Mill.jl and JsonGrinder.jl: automated differentiable feature extraction for learning from raw JSON data

Learning from raw data input, thus limiting the need for manual feature engineering, is one of the key components of many successful applications of machine learning methods. While machine learning problems are often formulated on data that naturally translate into a vector representation suitable for classifiers, there are data sources, for example in cybersecurity, that are naturally represented in diverse files with a unifying hierarchical structure, such as XML, JSON, and Protocol Buffers. Converting this data to vector (tensor) representation is generally done by manual feature engineering, which is laborious, lossy, and prone to human bias about the importance of particular features. Mill and JsonGrinder is a tandem of libraries, which fully automates the conversion. Starting with an arbitrary set of JSON samples, they create a differentiable machine learning model capable of infer from further JSON samples in their raw form.

stat.ML↗

arXiv2021

Multi-agent Reinforcement Learning in OpenSpiel: A Reproduction Report

In this report, we present results reproductions for several core algorithms implemented in the OpenSpiel framework for learning in games. The primary contribution of this work is a validation of OpenSpiel's re-implemented search and Reinforcement Learning algorithms against the results reported in their respective originating works. Additionally, we provide complete documentation of hyperparameters and source code required to reproduce these experiments easily and exactly.

cs.AI↗

arXiv2020

Discovering Imperfectly Observable Adversarial Actions using Anomaly Detection

Anomaly detection is a method for discovering unusual and suspicious behavior. In many real-world scenarios, the examined events can be directly linked to the actions of an adversary, such as attacks on computer networks or frauds in financial operations. While the defender wants to discover such malicious behavior, the attacker seeks to accomplish their goal (e.g., exfiltrating data) while avoiding the detection. To this end, anomaly detectors have been used in a game-theoretic framework that captures these goals of a two-player competition. We extend the existing models to more realistic settings by (1) allowing both players to have continuous action spaces and by assuming that (2) the defender cannot perfectly observe the action of the attacker. We propose two algorithms for solving such games -- a direct extension of existing algorithms based on discretizing the feature space and linear programming and the second algorithm based on constrained learning. Experiments show that both algorithms are applicable for cases with low feature space dimensions but the learning-based method produces less exploitable strategies and it is scalable to higher dimensions. Moreover, we use real-world data to compare our approaches with existing classifiers in a data-exfiltration scenario via the DNS channel. The results show that our models are significantly less exploitable by an informed attacker.

cs.LG↗

arXiv2020

Automated Construction of Bounded-Loss Imperfect-Recall Abstractions in Extensive-Form Games

Extensive-form games (EFGs) model finite sequential interactions between players. The amount of memory required to represent these games is the main bottleneck of algorithms for computing optimal strategies and the size of these strategies is often impractical for real-world applications. A common approach to tackle the memory bottleneck is to use information abstraction that removes parts of information available to players thus reducing the number of decision points in the game. However, existing information-abstraction techniques are either specific for a particular domain, they do not provide any quality guarantees, or they are applicable to very small subclasses of EFGs. We present domain-independent abstraction methods for creating imperfect recall abstractions in extensive-form games that allow computing strategies that are (near) optimal in the original game. To this end, we introduce two novel algorithms, FPIRA and CFR+IRA, based on fictitious play and counterfactual regret minimization. These algorithms can start with an arbitrary domain specific, or the coarsest possible, abstraction of the original game. The algorithms iteratively detect the missing information they require for computing a strategy for the abstract game that is (near) optimal in the original game. This information is then included back into the abstract game. Moreover, our algorithms are able to exploit imperfect-recall abstractions that allow players to forget even history of their own actions. However, the algorithms require traversing the complete unabstracted game tree. We experimentally show that our algorithms can closely approximate Nash equilibrium of large games using abstraction with as little as 0.9% of information sets of the original game. Moreover, the results suggest that memory savings increase with the increasing size of the original games.

cs.GT↗

arXiv2019

Monte Carlo Continual Resolving for Online Strategy Computation in Imperfect Information Games

Online game playing algorithms produce high-quality strategies with a fraction of memory and computation required by their offline alternatives. Continual Resolving (CR) is a recent theoretically sound approach to online game playing that has been used to outperform human professionals in poker. However, parts of the algorithm were specific to poker, which enjoys many properties not shared by other imperfect information games. We present a domain-independent formulation of CR applicable to any two-player zero-sum extensive-form games that works with an abstract resolving algorithm. We further describe and implement its Monte Carlo variant (MCCR) which uses Monte Carlo Counterfactual Regret Minimization (MCCFR) as a resolver. We prove the correctness of CR and show an $O(T^{-1/2})$-dependence of MCCR's exploitability on the computation time. Furthermore, we present an empirical comparison of MCCR with incremental tree building to Online Outcome Sampling and Information-set MCTS on several domains.

cs.GT↗

arXiv2017

Equilibrium Approximation Quality of Current No-Limit Poker Bots

Approximating a Nash equilibrium is currently the best performing approach for creating poker-playing programs. While for the simplest variants of the game, it is possible to evaluate the quality of the approximation by computing the value of the best response strategy, this is currently not computationally feasible for larger variants of the game, such as heads-up no-limit Texas hold'em. In this paper, we present a simple and computationally inexpensive Local Best Response method for computing an approximate lower bound on the value of the best response strategy. Using this method, we show that existing poker-playing programs, based on solving abstract games, are remarkably poor Nash equilibrium approximations.

cs.GT↗