The Experts below are selected from a list of 153 Experts worldwide ranked by ideXlab platform
Marek Kretowski - One of the best experts on this subject based on the ideXlab platform.
-
decision tree underfitting in mining of gene expression data an evolutionary multi test tree approach
Expert Systems With Applications, 2019Co-Authors: Marcin Czajkowski, Marek KretowskiAbstract:Abstract The problem of underfitting and overfitting in machine learning is often associated with a bias-variance trade-off. The underfitting most clearly manifests in the tree-based inducers when used to classify the gene expression data. To improve the generalization ability of decision trees, we are introducing an evolutionary, multi-test tree approach tailored to this Specific Application Domain. The general idea is to apply gene clusters of varying size, which consist of functionally related genes in each splitting rule. It is achieved by using a few simple tests that mimic each other’s predictions and built-in information about the discriminatory power of genes. The tendencies to underfit and overfit are limited by the multi-objective fitness function that minimizes tree error, split divergence and attribute costs. Evolutionary search for multi-tests in internal nodes, as well as the overall tree structure, is performed simultaneously. This novel approach called Evolutionary Multi-Test Tree (EMTTree) may bring far-reaching benefits to the Domain of molecular biology including biomarker discovery, finding new gene-gene interactions and high-quality prediction. Extensive experiments carried out on 35 publicly available gene expression datasets show that we managed to significantly improve the accuracy and stability of decision tree. Importantly, EMTTree does not substantially increase the overall complexity of the tree, so that the patterns in the predictive structures are kept comprehensible.
Marcin Czajkowski - One of the best experts on this subject based on the ideXlab platform.
-
decision tree underfitting in mining of gene expression data an evolutionary multi test tree approach
Expert Systems With Applications, 2019Co-Authors: Marcin Czajkowski, Marek KretowskiAbstract:Abstract The problem of underfitting and overfitting in machine learning is often associated with a bias-variance trade-off. The underfitting most clearly manifests in the tree-based inducers when used to classify the gene expression data. To improve the generalization ability of decision trees, we are introducing an evolutionary, multi-test tree approach tailored to this Specific Application Domain. The general idea is to apply gene clusters of varying size, which consist of functionally related genes in each splitting rule. It is achieved by using a few simple tests that mimic each other’s predictions and built-in information about the discriminatory power of genes. The tendencies to underfit and overfit are limited by the multi-objective fitness function that minimizes tree error, split divergence and attribute costs. Evolutionary search for multi-tests in internal nodes, as well as the overall tree structure, is performed simultaneously. This novel approach called Evolutionary Multi-Test Tree (EMTTree) may bring far-reaching benefits to the Domain of molecular biology including biomarker discovery, finding new gene-gene interactions and high-quality prediction. Extensive experiments carried out on 35 publicly available gene expression datasets show that we managed to significantly improve the accuracy and stability of decision tree. Importantly, EMTTree does not substantially increase the overall complexity of the tree, so that the patterns in the predictive structures are kept comprehensible.
Hyun Soo Kim - One of the best experts on this subject based on the ideXlab platform.
-
a causal knowledge driven inference engine for expert system
Hawaii International Conference on System Sciences, 1998Co-Authors: Kun Chang Lee, Hyun Soo KimAbstract:A wide variety of knowledge acquisition methods exist for conventional knowledge types such as production rules, semantic knowledge, etc. However, the need for causal knowledge acquisition has not been stressed in the expert systems field. The objectives of this paper are to: suggest a causal knowledge acquisition process; and investigate the causal knowledge-based inference process. FCM (Fuzzy Cognitive Map), a fuzzy signed digraph with causal relationships between concept variables found in a Specific Application Domain, is used for the causal knowledge acquisition. Although FCM has plenty of generic properties for causal knowledge acquisition, it needs some theoretical improvement for acquiring more refined causal knowledge. In this sense, we refine fuzzy implications of FCM by proposing a fuzzy causal relationship and fuzzy partially causal relationship. To test the validity of our proposed approach, we prototype a causal knowledge-driven inference engine named CAKES and then experiment with illustrative examples.
-
a causal knowledge driven inference engine for expert system
Journal of The Korean Institute of Intelligent Systems, 1998Co-Authors: Kun Chang Lee, Hyun Soo KimAbstract:Although many methods of knowledge acquisition has been developed in the exper systems field, such a need form causal knowledge acquisition hs not been stressed relatively. In this respect, this paper is aimed at suggesting a causal knowledge acquisition process, and then investigate the causal knowledge-based inference process. A vehicle for causal knowledge acquisition is FCM (Fuzzy Cognitive Map), a fuzzy signed digraph with causal relationships between concept variables found in a Specific Application Domain. Although FCM has a plenty of generic properties for causal knowledge acquisition, it needs some theoretical improvement for acquiring a more refined causal knowledge. In this sense, we refine fuzzy implications of FCM by proposing fuzzy causal relationship and fuzzy partially causal relationship. To test the validity of our proposed approach, we prototyped a causal knowledge-driven inference engine named CAKES and then experimented with some illustrative examples.
Samuel Pierre - One of the best experts on this subject based on the ideXlab platform.
-
mobile agents and their use for information retrieval a brief overview and an elaborate case study
IEEE Network, 2002Co-Authors: Roch Glitho, Edgar Olougouna, Samuel PierreAbstract:Mobile agents emerged in the mid-1990s, and have raised considerable interest in the research community. The proponents associate several benefits with their use. However, there are still very few quantitative measurements to back the claimed benefits. This article is devoted to mobile agents and their use for information retrieval. We provide a brief overview and an elaborate case study. The overview introduces the concept of mobile agent, enumerates the claimed benefits, and reviews the hindrances to widescale deployment. It also discusses the state of the art of mobile-agent-based information retrieval, including the very few quantitative studies that exist. Our case study is on information retrieval from electronic calendars for multiparty event scheduling. Many events require the participation of several parties. Prior knowledge of the date when most (if not all) targeted participants are available is often a prerequisite for scheduling them. However, identifying this date can easily turn into a nightmare, especially when the number of targeted participants is large. Nowadays, electronic agendas (e.g., MS Outlook) are stored on servers. An Application can access them, retrieve information on the availability of the targeted participants, and derive the date from the information. In the case study, a mobile agent is dispatched in the network, instead of retrieving the information using the client/server paradigm. The agent visits the servers, accesses the agendas, retrieves the information, and identifies the date. Finding a date suitable for several potential participants may require the rescheduling of some events that have been previously arranged by some participants. We propose the use of agents that act as the personal agents of the participants for the negotiation inherent to this rescheduling. The measurements we have made indicate clearly that the mobile-agent-based approach outperforms its client/server counterpart even when the latter is optimized. These results can easily be transposed to most information retrieval Applications, and demonstrate, for this Specific Application Domain, the performance benefit associated with mobile agents. We now dispatch a single agent in the network. In the future, we will dispatch several agents.
Sergei Sawitzki - One of the best experts on this subject based on the ideXlab platform.
-
a heterogeneous architecture template for Application Domain Specific reconfigurable logic
2015 Austrian Workshop on Microelectronics, 2015Co-Authors: Timm Bostelmann, Sergei SawitzkiAbstract:In this paper we present a heterogeneous architecture template for reconfigurable logic and its proposed design flow. From this template which we also call a meta-architecture a set of reconfigurable architectures can be derived and optimized towards a Specific Application Domain. The meta-architecture is very flexible and allows a wide and deep exploration of the architecture-design-space. Our goal is to provide the user with the ability to tailor a reconfigurable architecture to a Specific Application Domain, for example communication or image processing. The function blocks and routing structures can be modified as well as the global structure of the design. Thereby the disadvantages of very flexible, universal structures like FPGAs (inefficient resource usage and high communication overhead) can be diminished. At the same time their advantages (short time to market and low non-recurring engineering costs) can be kept. We show the degrees of freedom the proposed meta-architecture offers and discuss the corresponding trade-offs. Furthermore we show in which ways the architecture can be optimized for an Application Domain and why it is more flexible then island-style structures. We also present the proposed design flow and discuss the implementation of its stages. Finally we present the status of this work in progress and give a prospect to the planned future work.
-
a conceptual toolchain for an Application Domain Specific reconfigurable logic architecture
Reconfigurable Computing and FPGAs, 2014Co-Authors: Timm Bostelmann, Sergei SawitzkiAbstract:In this paper we present a concept of a reconfigurable logic toolchain. The specialty of this toolchain is the highly configurable architecture design. The goal is to provide the designer with the ability to suit the architecture of the reconfigurable logic to a Specific Application Domain, for example communication or image processing. Thereby the disadvantages of very flexible, universal structures like FPGAs (inefficient resource usage and high communication overhead) can be diminished. At the same time their advantages (short time to market and low non-recurring engineering costs) can be kept. To achieve this a graphical architecture editor allows the user to adapt the global design structure as well as the detailed implementation of the logic cells. An analysis tool reports how frequently the different logic and routing resources of the described architecture are utilized by a given set of Applications, to allow an optimization towards a Specific Application Domain. We show the degrees of freedom the envisioned toolchain offers and discuss the corresponding trade-offs. Furthermore we show which steps of the development toolchain have to be adapted to the needs of such a flexible architecture and how this can be done. Finally we present the status of this work in progress and give a prospect to the planned future work.
