Proceedings of the Genetic and Evolutionary Computation Conference Companion
This paper explains the process of integrating ACS2 algorithm with the standardised framework for... more This paper explains the process of integrating ACS2 algorithm with the standardised framework for comparing reinforcement learning tasks-OpenAI Gym. The new Python library is introduced alongside with standard environments derived from LCS literature. Typical use cases enabling quick evaluation of di erent research problems are described. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
Proceedings of the Genetic and Evolutionary Computation Conference Companion
The paper describes the first attempts toward designing and evaluating Anticipatory Classifier Sy... more The paper describes the first attempts toward designing and evaluating Anticipatory Classifier System ACS2 in conjunction with Experience Replay (ER). Promising results verified by statistical tests are obtained both on single-and multi-step problems, albeit limited to deterministic and discrete tasks. The analysis indicates that ACS2 using memorized experiences has the potential for significant improvements in learning efficiency and knowledge generality. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
Proceedings of the Genetic and Evolutionary Computation Conference Companion
This paper explains the process of integrating ACS2 algorithm with the standardised framework for... more This paper explains the process of integrating ACS2 algorithm with the standardised framework for comparing reinforcement learning tasks-OpenAI Gym. The new Python library is introduced alongside with standard environments derived from LCS literature. Typical use cases enabling quick evaluation of di erent research problems are described. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
Proceedings of the Genetic and Evolutionary Computation Conference Companion
The paper describes the first attempts toward designing and evaluating Anticipatory Classifier Sy... more The paper describes the first attempts toward designing and evaluating Anticipatory Classifier System ACS2 in conjunction with Experience Replay (ER). Promising results verified by statistical tests are obtained both on single-and multi-step problems, albeit limited to deterministic and discrete tasks. The analysis indicates that ACS2 using memorized experiences has the potential for significant improvements in learning efficiency and knowledge generality. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
WSEAS TRANSACTIONS on SYSTEMS archive, Oct 1, 2008
The paper introduces an algorithmic improvement to IFRAIS, an existing Artificial Immune System m... more The paper introduces an algorithmic improvement to IFRAIS, an existing Artificial Immune System method for fuzzy rule mining. The improvement presented consists of using rule buffering during the computation of fitness of rules. This is achieved using a hash table. The improved method has been tested against two different fitness functions and various data sets. Experimental results show improvements in computing times in the order of 3 to 10 times maintaining same levels of accuracy.
WSEAS TRANSACTIONS on SYSTEMS archive, Oct 1, 2008
The paper introduces an algorithmic improvement to IFRAIS, an existing Artificial Immune System m... more The paper introduces an algorithmic improvement to IFRAIS, an existing Artificial Immune System method for fuzzy rule mining. The improvement presented consists of using rule buffering during the computation of fitness of rules. This is achieved using a hash table. The improved method has been tested against two different fitness functions and various data sets. Experimental results show improvements in computing times in the order of 3 to 10 times maintaining same levels of accuracy.
Extended abstract of the 14th International Conference on Grammatical Inference ICGI’18, Septembe... more Extended abstract of the 14th International Conference on Grammatical Inference ICGI’18, September 5–7, 2018, Wrocław, Poland. Abstract A modified inside-outside algorithm to estimate probabilistic parameters over implicit positive and negative evidence was proposed. We have demonstrated that a contrastive estimation based method significantly outperforms a standard inside-outside algorithm in terms of Specificity, without any loss of Sensitivity.
Extended abstract of the 14th International Conference on Grammatical Inference ICGI’18, Septembe... more Extended abstract of the 14th International Conference on Grammatical Inference ICGI’18, September 5–7, 2018, Wrocław, Poland. Abstract A modified inside-outside algorithm to estimate probabilistic parameters over implicit positive and negative evidence was proposed. We have demonstrated that a contrastive estimation based method significantly outperforms a standard inside-outside algorithm in terms of Specificity, without any loss of Sensitivity.
This proceedings contains the works that have been presented at the 14th International Conference... more This proceedings contains the works that have been presented at the 14th International Conference on Grammatical Inference (ICGI), held in Wroc law, Poland, from September 5 through September 7, 2018. Out of the 17 full papers, 11 were accepted for publication in the proceedings and presentation at the conference. In addition to these works, four extended abstracts were accepted for a short presentation at the conference and two extended abstracts—for a poster session. The extended abstracts are not included in the proceedings, but can be found online at http://icgi2018.pwr.edu.pl. The proceedings contains a diverse range of topics in grammatical inference, such as: new ideas in automata learning, inferring context-free grammars, learning weighted automata and other soft classifiers, as well as the application of these methods in solving hard tasks and modeling complex systems. This year the conference was held in collocation with the annual symposium organized by the Polish Bioinfo...
This proceedings contains the works that have been presented at the 14th International Conference... more This proceedings contains the works that have been presented at the 14th International Conference on Grammatical Inference (ICGI), held in Wroc law, Poland, from September 5 through September 7, 2018. Out of the 17 full papers, 11 were accepted for publication in the proceedings and presentation at the conference. In addition to these works, four extended abstracts were accepted for a short presentation at the conference and two extended abstracts—for a poster session. The extended abstracts are not included in the proceedings, but can be found online at http://icgi2018.pwr.edu.pl. The proceedings contains a diverse range of topics in grammatical inference, such as: new ideas in automata learning, inferring context-free grammars, learning weighted automata and other soft classifiers, as well as the application of these methods in solving hard tasks and modeling complex systems. This year the conference was held in collocation with the annual symposium organized by the Polish Bioinfo...
This paper introduces a new tool for context-sensitive grammar inference. The source code and lib... more This paper introduces a new tool for context-sensitive grammar inference. The source code and library are publicly available via GitHub and NuGet repositories. The article describes the implemented algorithm, input parameters, and the produced output grammar. In addition, the paper contains several use-cases. The described library is written in F#, hence it can be used in any .NET Framework language (F#, C#, C++/CLI, Visual Basic, and J#) and run under the control of varied operating systems.
This paper introduces a new tool for context-sensitive grammar inference. The source code and lib... more This paper introduces a new tool for context-sensitive grammar inference. The source code and library are publicly available via GitHub and NuGet repositories. The article describes the implemented algorithm, input parameters, and the produced output grammar. In addition, the paper contains several use-cases. The described library is written in F#, hence it can be used in any .NET Framework language (F#, C#, C++/CLI, Visual Basic, and J#) and run under the control of varied operating systems.
Grammatical inference is a machine learning area, whose fundamentals are built around learning se... more Grammatical inference is a machine learning area, whose fundamentals are built around learning sets. At present, real-life data and examples from manually crafted grammars are used to test their learning performance. This paper aims to present a method of generating artificial context-free grammars with their optimal learning sets, which could be successfully applied as a benchmarking tool for empirical grammar inference methods.
Grammatical inference is a machine learning area, whose fundamentals are built around learning se... more Grammatical inference is a machine learning area, whose fundamentals are built around learning sets. At present, real-life data and examples from manually crafted grammars are used to test their learning performance. This paper aims to present a method of generating artificial context-free grammars with their optimal learning sets, which could be successfully applied as a benchmarking tool for empirical grammar inference methods.
Context-free and context-sensitive formal grammars are often regarded as more appropriate to mode... more Context-free and context-sensitive formal grammars are often regarded as more appropriate to model proteins than regular level models such as finite state automata and Hidden Markov Models. In theory, the claim is well founded in the fact that many biologically relevant interactions between residues of protein sequences have a character of nested or crossed dependencies. In practice, there is hardly any evidence that grammars of higher expressiveness have an edge over old good HMMs in typical applications including recognition and classification of protein sequences. This is in contrast to RNA modeling, where CFG power some of the most successful tools. There have been proposed several explanations of this phenomenon. On the biology side, one difficulty is that interactions in proteins are often less specific and more "collective" in comparison to RNA. On the modeling side, a difficulty is the larger alphabet which combined with high complexity of CF and CS grammars impose...
Proceedings of the Genetic and Evolutionary Computation Conference Companion, 2019
The paper describes the irst attempts toward designing and evaluating anticipatory classiier syst... more The paper describes the irst attempts toward designing and evaluating anticipatory classiier systems working in a real-valued input domain using interval predicates representation. Promising results are obtained by testing two environments-real-valued multiplexer and checkerboard from the classical XCSR problem domain. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion, 2020
One way of dealing with the real-valued input signal is to discretize it. This might influence th... more One way of dealing with the real-valued input signal is to discretize it. This might influence the process of learning the environmental model by the ACS2 agent. A more sophisticated method of selecting action can be applied to increase the speed of gaining knowledge by determining the most valuable regions of the input-space. This paper compares four ACS2 biasing exploration techniques applied across four real-valued environments. A new class of benchmark problem (inverted pendulum) and an agent modification-Optimistic Initial Quality (OIQ) are introduced for ACS2 both with promising outcomes. CCS CONCEPTS • Computing methodologies → Rule learning; • Software and its engineering;
Proceedings of the 12th International Conference on Agents and Artificial Intelligence, 2020
In this paper, we address the problem of inducing (weighted) context-free grammar (WCFG) on data ... more In this paper, we address the problem of inducing (weighted) context-free grammar (WCFG) on data given. The induction is performed by using a new model of grammatical inference, i.e., weighted Grammar-based Classifier System (wGCS). wGCS derives from learning classifier systems and searches grammar structure using a genetic algorithm and covering. Weights of rules are estimated by using a novelty Inside-Outside Contrastive Estimation algorithm. The proposed method employs direct negative evidence and learns WCFG both form positive and negative samples. Results of experiments on three synthetic context-free languages show that wGCS is competitive with other statistical-based method for unsupervised CFG learning.
Uploads
Papers by Olgierd Unold