Tomohiro Takagi

Multilayered reasoning by means of conceptual fuzzy sets

The real world consists of instances of events and continuous numeric values, while people represent and process their knowledge in terms of symbols. Fuzzy sets provide a strong notation connecting the symbolic representation to the real world. In previously proposed Conceptual Fuzzy Sets (CFS), the meaning of a concept is represented by the distribution of activations of labels in a bidirectional associative memory. In particular, a multilayered structured CFS represents the meaning of the same concept as it is used in various expressions in each layer. The propagation of activations corresponds to reasoning. Therefore, we propose a multilayered reasoning method associated to a multilayered structured CFS, which has the following features: (1) capable of simultaneous symbolic and quantitative processing, (2) capable of simultaneous top‐down and bottom‐up processing. The effectiveness of the proposed method is illustrated by practical examples of decision regarding the amount of steering in the task of parking a car, and recognition of facial expressions for an image understanding system.

Conceptual fuzzy sets as a meaning representation and their inductive construction

Fuzzy sets provide a strong notation for representing real world concepts which are essentially vague. However they have problems caused by the restriction of numerical membership functions, restriction of logical expression, lack of context dependency, etc. These problems relate to the representation of the meaning of a concept. In this article, we propose Conceptual Fuzzy Sets (CFS), a new type of fuzzy sets which conform to Wittgensteins ideas (Philosophical Investigations, Basil Blackwell, Oxford, 1953) on concept meaning. A CFS is realized as an associative memory, combining a long‐term memory and a short‐term memory thus reducing the complexity of knowledge representation. In addition to solving the above problems CFS provide simple formula for knowledge representation and the procedure to use this knowledge. We introduce an inductive method for constructing CFS based on neural network learning. the effectiveness of CFS and of the learning method is illustrated through their application to the recognition of facial expressions.

Conceptual matching and its applications to selection of TV programs and BGMs

We propose a method for representing common concepts by fusing conceptual fuzzy sets (CFSs) with ontology. We also describe a method of conceptual matching for retrieving information that meets a users intentions. The matching is done based on the region of the ontological structure covered by the CFS representing the users intention as well as on the original keywords. We apply the proposed conceptual-matching method to two types of agents. One agent recommends TV programs to watch. The recommended programs have EPGs (electronic program guides) similar to those of previously watched programs or that contain words matching the learning data. Practical examples demonstrate that the proposed system recommend TV programs matching the users tastes. The other selects music matching the tone of a text composition.

Recognition of facial expressions using conceptual fuzzy sets

A facial expression is a vague concept that is difficult to describe explicitly. Conceptual fuzzy sets (CFSs), which have the ability to explicitly represent vague concepts, are realized using bidirectional associative memories, and a multilayer structured CFS represents the meaning of a concept by various expressions in each layer. Multilayered reasoning in CFS has the capability of simultaneous abstract and concrete representation and of simultaneous top-down and bottom-up processing. CFS has been applied to the recognition of facial expressions and shown to achieve context-sensitive recognition.<<ETX>>

Query expansion using conceptual fuzzy sets for search engine

We propose a search engine which conceptually matches input keywords and web pages. The conceptual matching is realized by context-dependent keyword expansion using conceptual fuzzy sets. First, we show the necessity and the problems of applying fuzzy sets to information retrieval. Next, we introduce the usefulness of conceptual fuzzy sets in overcoming those problems, and propose the realization of conceptual fuzzy sets using Hopfield networks. We also propose the architecture of the search engine which can execute conceptual matching dealing with context-dependent word ambiguity. Finally, we evaluate our proposed method through two simulations of retrieving actual web pages, and compare the proposed method with the ordinary TF-IDF method. We show that our method can correlate seemingly unrelated input keywords and produce matching Web pages, whereas the TF-IDF method cannot.

Intelligent control of a flying vehicle using fuzzy associative memory system

The authors propose a flying vehicle intelligent control system which simulates the pilots operation knowledge and training steps. To simulate the operation knowledge and training steps, a fuzzy associative memory system called FAMOUS was used. FAMOUS uses associative memory neural networks which represent fuzzy knowledge. There are two types of operation knowledge; dynamic fuzzy knowledge, for example the circular flight operation pattern; and static fuzzy knowledge, for example the hovering operation model corresponding to each flying condition. FAMOUS represents both dynamic and static fuzzy knowledge using its hierarchical knowledge representation. The implementation of the intelligent control system using FAMOUS and the realization of hovering flight and circular flight are discussed.<<ETX>>

Two-degree-of-freedom fuzzy model using associative memories and its applications

Abstract We propose a model for a physical plant system. This model simulates the steps of an experts training, and that simulates an experts knowledge, about steady state operation of a plant, about the plants dynamic transitions knowledge, and the training steps. To represent this model, we use an associative memory system called fuzzy associative memory organizing units system (FAMOUS) to structure two types of knowledge: 1. (1) static fuzzy knowledge (SFK), i.e., about operations corresponding to each operational condition, and 2. (2) dynamic fuzzy knowledge (DFK), i.e., about dynamic state-transition patterns generation under all conditions. We call this model using two types of fuzzy knowledge a two-degree-of-freedom fuzzy model. It is difficult to use if-then rules to represent the featured phenomenon (i.e., a series of dynamic state-transition patterns together with their characteristic fluctuations) because the rule representation is too complex to be acquired from experts. The two-degree-of-freedom fuzzy model, however, can represent the featured phenomena by using a combination of compact SFK and DFK similar to the knowledge acquired through experience by human beings. Fuzzy knowledge from experts is initially put into FAMOUS, and then refined according to the experts ideal operations and the plants states by using a learning algorithm. After learning the fuzzy knowledge, the uncertain knowledge is more desirable for representing the featured phenomena than before learning. The two-degree-of-freedom fuzzy model uses associative memories to achieve operation and prediction close to those of human beings. In addition, application examples are reported: the flight control of a small four-propeller flying vehicle (similar to a helicopter) and the smooth running of a pump station of a sewage treatment plant. We also give an outline of the fuzzy-type associative memory and the two-degree-of-freedom fuzzy model, the extraction and refinement of knowledge for stabilizing a physical plant, and for a series of dynamic state-transition patterns together with the characteristic fluctuations.

Realization of sound-scape agent by the fusion of conceptual fuzzy sets and ontology

We propose a method for representing common concepts by fusing conceptual fuzzy sets (CFS) with ontology. We also describe a method of conceptual matching for retrieving information that meets a users intentions. The matching is done based on the region of the ontological structure covered by the CFS representing the users intention as well as on the original keywords. We apply the proposed conceptual-matching method to an agent, which selects music matching the tone of a text composition. Selection of music for two e-mail notes shows that the proposed system can select music that matches human feelings.

Recursive-Rule Extraction Algorithm With J48graft And Applications To Generating Credit Scores

Abstract The purpose of this study was to generate more concise rule extraction from the Recursive-Rule Extraction (Re-RX) algorithm by replacing the C4.5 program currently employed in Re-RX with the J48graft algorithm. Experiments were subsequently conducted to determine rules for six different two-class mixed datasets having discrete and continuous attributes and to compare the resulting accuracy, comprehensibility and conciseness. When working with the CARD1, CARD2, CARD3, German, Bene1 and Bene2 datasets, Re-RX with J48graft provided more concise rules than the original Re-RX algorithm. The use of Re-RX with J48graft resulted in 43.2%, 37% and 21% reductions in rules in the case of the German, Bene1 and Bene2 datasets compared to Re-RX. Furthermore, the Re-RX with J48graft showed 8.87% better accuracy than the Re-RX algorithm for the German dataset. These results confirm that the application of Re-RX in conjunction with J48graft has the capacity to facilitate migration from existing data systems toward new concise analytic systems and Big Data.

Conceptual fuzzy sets-based navigation system for Yahoo!

In this paper, we propose a menu navigation system which conceptually matches input keywords and paths. For conceptual matching, we use conceptual fuzzy sets (CFSs) based on radial basis function (RBF) networks. In a CFS, the meaning of a concept is represented by the distribution of the activation values of the other concepts. To expand input keywords the propagation of activation values is carried out recursively. The proposed system recommends users paths to appropriate categories. We use 3D user interface to navigate users.