Research (CBII, Kansai University)

http://www.kansai-u.ac.jp/Kokusai/e-bulletin/archive/14.php
The brain intelligence model is to observe human brain signals and behavior patterns, and to realize the same behavior with algorithms and programs. In our laboratory, we utilize neural network, fuzzy system, genetic algorithm, and machine learning as models of artificial intelligence and soft computing.

Research

Collaborative Research and Development of OMRON's Table Tennis Robot "FORPHEUS"
https://www.omron.com/global/en/technology/information/forpheus/
The name FORPHEUS comes from "Future OMRON Robotics technology for Exploring Possibility of Harmonized aUtomation with Sinic theoretics". This term is a coined word embodying OMRON's robot technologies based on its unique future prediction theory "SINIC theoretics". Hayashi Laboratory, Faculty of Informatics of Kansai University conducts collaborative research and development of OMRON's table tennis robot "FORPHEUS" under the joint participation of the Table Tennis Club of Kansai University.

Collaborative Research and Development of OMRON's Table Tennis Robot "FORPHEUS"

https://www.omron.com/global/en/technology/information/forpheus/

The name FORPHEUS comes from "Future OMRON Robotics technology for Exploring Possibility of Harmonized aUtomation with Sinic theoretics". This term is a coined word embodying OMRON's robot technologies based on its unique future prediction theory "SINIC theoretics". Hayashi Laboratory, Faculty of Informatics of Kansai University conducts collaborative research and development of OMRON's table tennis robot "FORPHEUS" under the joint participation of the Table Tennis Club of Kansai University.

pdi-Bagging: A Proposal of Bagging-type Ensemble Method Generating Virtual Data
https://www.cbii.kutc.kansai-u.ac.jp/conference/229.pdf
For pattern classification problems, there is ensemble learning method that identifies multiple weak classifiers by the learning data and combines them together to improve the discrimination rate of testing data. We have already proposed pdi-Bagging (Possibilistic Data Interpolation-Bagging) which improves the discrimination rate of testing data by adding virtually generated data to learning data. In this paper, we propose a new method to specify the generation area of virtual data and change the generation class of virtual data. As a result, the discriminant accuracy is improved since five new bagging methods which generate virtual data around correct discrimination data and error discrimination data are formulated, and the class of virtual data is determined with the proposed new evaluation index in multidimensional space. We formulate a new pdi-Bagging algorithm, and discuss the usefulness of the proposed method using numerical examples.

pdi-Bagging: A Proposal of Bagging-type Ensemble Method Generating Virtual Data

https://www.cbii.kutc.kansai-u.ac.jp/conference/229.pdf

For pattern classification problems, there is ensemble learning method that identifies multiple weak classifiers by the learning data and combines them together to improve the discrimination rate of testing data. We have already proposed pdi-Bagging (Possibilistic Data Interpolation-Bagging) which improves the discrimination rate of testing data by adding virtually generated data to learning data. In this paper, we propose a new method to specify the generation area of virtual data and change the generation class of virtual data. As a result, the discriminant accuracy is improved since five new bagging methods which generate virtual data around correct discrimination data and error discrimination data are formulated, and the class of virtual data is determined with the proposed new evaluation index in multidimensional space. We formulate a new pdi-Bagging algorithm, and discuss the usefulness of the proposed method using numerical examples.

Time-Series Data Analysis Using Sliding Window Based SVD for Motion Evaluation
https://www.cbii.kutc.kansai-u.ac.jp/paper/46.pdf
A method based on singular value decomposition (SVD) is proposed for extracting features from motion time-series data observed with various sensing systems. Matrices consisting of the sliding window(SW) subsets of time-series data are decomposed, yielding singular vectors as the patterns of the motion, and the singular values as a scalar, by which the corresponding singular vectors describe the matrices. The sliding window based singular value decomposition was applied to analyze acceleration during walking. Three levels of walking difficulty were simulated by restricting the right knee joint in the measurement. The accelerations of the middles of the shanks and the back of the waist were measured and normalized before the SW-SVD was performed.The results showed that the first singular values inferred from the acceleration data of the restricted side (the right shank) significantly related to the increase of the restriction among all the subjects while there were no common trends in the singular values of the left shank and the waist. The SW-SVD was suggested to be a reliable method to evaluate walking disability. Furthermore, a 2D visualization tool is proposed to provide intuitive information about walking difficulty which can be used in walking rehabilitation to monitor recovery.

Time-Series Data Analysis Using Sliding Window Based SVD for Motion Evaluation

https://www.cbii.kutc.kansai-u.ac.jp/paper/46.pdf

A method based on singular value decomposition (SVD) is proposed for extracting features from motion time-series data observed with various sensing systems. Matrices consisting of the sliding window(SW) subsets of time-series data are decomposed, yielding singular vectors as the patterns of the motion, and the singular values as a scalar, by which the corresponding singular vectors describe the matrices. The sliding window based singular value decomposition was applied to analyze acceleration during walking. Three levels of walking difficulty were simulated by restricting the right knee joint in the measurement. The accelerations of the middles of the shanks and the back of the waist were measured and normalized before the SW-SVD was performed.The results showed that the first singular values inferred from the acceleration data of the restricted side (the right shank) significantly related to the increase of the restriction among all the subjects while there were no common trends in the singular values of the left shank and the waist. The SW-SVD was suggested to be a reliable method to evaluate walking disability. Furthermore, a 2D visualization tool is proposed to provide intuitive information about walking difficulty which can be used in walking rehabilitation to monitor recovery.

Extraction of Knowledge from the Topographic Attentive Mapping Network and its Application in Skill Analysis of Table Tennis
https://www.cbii.kutc.kansai-u.ac.jp/paper/45.pdf
The Topographic Attentive Mapping (TAM) network is a biologically-inspired classifier that bears similarities to the human visual system. In case of wrong classification during training, an attentional top-down signal modulates synaptic weights in intermediate layers to reduce the difference between the desired output and the classifier’s output. When used in a TAM network, the proposed pruning algorithm improves classification accuracy and allows extracting knowledge as represented by the network structure. In this paper, sport technique evaluation of motion analysis modelled by the TAM network was discussed. The trajectory pattern of forehand strokes of table tennis players was analyzed with nine sensor markers attached to the right upper arm of players. With the TAM network, input attributes and technique rules were extracted in order to classify the skill level of players of table tennis from the sensor data. In addition, differences between the elite player, middle level player and beginner were clarified; furthermore, we discussed how to improve skills specific to table tennis from the view of data analysis.

Extraction of Knowledge from the Topographic Attentive Mapping Network and its Application in Skill Analysis of Table Tennis

https://www.cbii.kutc.kansai-u.ac.jp/paper/45.pdf

The Topographic Attentive Mapping (TAM) network is a biologically-inspired classifier that bears similarities to the human visual system. In case of wrong classification during training, an attentional top-down signal modulates synaptic weights in intermediate layers to reduce the difference between the desired output and the classifier’s output. When used in a TAM network, the proposed pruning algorithm improves classification accuracy and allows extracting knowledge as represented by the network structure. In this paper, sport technique evaluation of motion analysis modelled by the TAM network was discussed. The trajectory pattern of forehand strokes of table tennis players was analyzed with nine sensor markers attached to the right upper arm of players. With the TAM network, input attributes and technique rules were extracted in order to classify the skill level of players of table tennis from the sensor data. In addition, differences between the elite player, middle level player and beginner were clarified; furthermore, we discussed how to improve skills specific to table tennis from the view of data analysis.

Vitroid - The Robot System with an Interface Between a Living Neuronal Network and Outer World
https://www.cbii.kutc.kansai-u.ac.jp/paper/35.pdf
We have developed a neuro-robot-hybrid system using a living neuronal network and a miniature moving robot. The living network of rat hippocampal neurons can distinguish patterns of action potentials evoked by different inputs, suggesting that a cultured neuronal network can represent particular states as symbols. We used a Khepera II robot and a robot made using a LEGO mindstorm NXT kit to interface with a living neuronal network and the outer world. We call the system ‘vitroid’. Vitroid has living neurons, a robot body, and direct coupling controllers to interface the neurons with the robot. Vitroid was able to perform obstacle avoidance behaviour with premised control rule sets.

Vitroid - The Robot System with an Interface Between a Living Neuronal Network and Outer World

https://www.cbii.kutc.kansai-u.ac.jp/paper/35.pdf

We have developed a neuro-robot-hybrid system using a living neuronal network and a miniature moving robot. The living network of rat hippocampal neurons can distinguish patterns of action potentials evoked by different inputs, suggesting that a cultured neuronal network can represent particular states as symbols. We used a Khepera II robot and a robot made using a LEGO mindstorm NXT kit to interface with a living neuronal network and the outer world. We call the system ‘vitroid’. Vitroid has living neurons, a robot body, and direct coupling controllers to interface the neurons with the robot. Vitroid was able to perform obstacle avoidance behaviour with premised control rule sets.

Description of Activity of Living Neuronal Network by Fuzzy Bio-Indicator
https://www.cbii.kutc.kansai-u.ac.jp/conference/187.pdf
The culture dish describes the small fundamental world resembling human brain function. Multi-site recording system for extracellular action potentials is used for recording the activity of living neuronal networks. The living neuronal network is able to express several patterns independently, and that's meaning that it has fundamental mechanisms for intelligent information processing. In this paper, we propose a model to analyse logicality of signals and connectivity of electrodes in a culture dish of rat hippocampal neurons. We call it ``fuzzy bio-indicator''. This indicator is a kind of mapping methods to show logicality and connectivity of pulse frequency from active potential of neuronal network. We try to analyze the dynamics of action potentials of neuronal networks by the fuzzy bio-indicator, and identify the logicality and connectivity of neuronal networks through the indicator. We show here the usefulness of fuzzy bio-indicator through numerical examples and action potential detected from the culture neuronal network.

Description of Activity of Living Neuronal Network by Fuzzy Bio-Indicator

https://www.cbii.kutc.kansai-u.ac.jp/conference/187.pdf

The culture dish describes the small fundamental world resembling human brain function. Multi-site recording system for extracellular action potentials is used for recording the activity of living neuronal networks. The living neuronal network is able to express several patterns independently, and that's meaning that it has fundamental mechanisms for intelligent information processing. In this paper, we propose a model to analyse logicality of signals and connectivity of electrodes in a culture dish of rat hippocampal neurons. We call it ``fuzzy bio-indicator''. This indicator is a kind of mapping methods to show logicality and connectivity of pulse frequency from active potential of neuronal network. We try to analyze the dynamics of action potentials of neuronal networks by the fuzzy bio-indicator, and identify the logicality and connectivity of neuronal networks through the indicator. We show here the usefulness of fuzzy bio-indicator through numerical examples and action potential detected from the culture neuronal network.

A Proposal for Applying pdi-Boosting to Brain-Computer Interfaces
https://www.cbii.kutc.kansai-u.ac.jp/conference/168.pdf
Brain-computer interface (BCI) and brain-machine interface (BMI) technologies have recently entered the research limelight. In many such systems, external computers and machines are controlled by brain activity signals measured using near-infrared spectroscopy (NIRS) or electroencephalograph (EEG) devices. In this paper, we propose a novel boosting algorithm for BCI using a probabilistic data interpolation scheme. In our model, interpolated data is generated around classification errors using a probability distribution function, as opposed to conventional AdaBoost which increases weights corresponding to the misclassified examples. By using the interpolated data, the discriminated boundary is shown to control the external machine effectively. We verify our boosting method with an experiment in which NIRS data is obtained from subjects performing a basic arithmetic task, and discuss the results.

A Proposal for Applying pdi-Boosting to Brain-Computer Interfaces

https://www.cbii.kutc.kansai-u.ac.jp/conference/168.pdf

Brain-computer interface (BCI) and brain-machine interface (BMI) technologies have recently entered the research limelight. In many such systems, external computers and machines are controlled by brain activity signals measured using near-infrared spectroscopy (NIRS) or electroencephalograph (EEG) devices. In this paper, we propose a novel boosting algorithm for BCI using a probabilistic data interpolation scheme. In our model, interpolated data is generated around classification errors using a probability distribution function, as opposed to conventional AdaBoost which increases weights corresponding to the misclassified examples. By using the interpolated data, the discriminated boundary is shown to control the external machine effectively. We verify our boosting method with an experiment in which NIRS data is obtained from subjects performing a basic arithmetic task, and discuss the results.