|題 目||「Constructive Neural Networks and Applications」|
|概 要|| The theory of Neural Networks (NNs) has witnessed a striking progress in the past decades. The basic issues, such as determining the structure and size of the network, and developing efficient training/learning strategies have always been major issues. This talk is mainly focused on constructive neural networks and their applications to regression, image compression, pattern recognition and solar radiation prediction/forecast problems.
First, two new strategies are introduced for a constructive One-Hidden-Layer Feedforward NN (OHL-FFNN) that grows from a small initial network with a few hidden units to one that has sufficient number of hidden units as required by the underlying mapping problem. The first strategy denoted as error scaling is designed to improve the training efficiency and generalization performance of the OHL-FFNN. The second strategy is a pruning criterion that produces a smaller network while not degrading the generalization capability of the network.
Second, a novel strategy at the structure level adaptation is developed for constructing multi-hidden-layer FFNNs. By utilizing the developed scheme, an FFNN is obtained that has sufficient number of hidden layers and hidden units that are required by the complexity of the mapping being considered.
Third, a new constructive OHL-FFNN at the functional level adaptation is developed. According to this scheme, each hidden unit uses a polynomial as its activation function that is different from those of the other units. This permits the growing network to employ different activation functions so that the network would be able to represent and capture the underlying map more efficiently as compared to the fixed activation function networks.
Finally, the developed constructive algorithms are applied to regression, still and moving image compression, facial expression recognition, and solar radiation prediction/forecast problems. It has been shown through extensive simulations that all the developed techniques and networks produce very promising results.
|講 師||Department of Electrical and Computer Engineering, Concordia University, Montreal, Canada, Dr. Liying Ma 氏|
|申込・問合せ||広島大学 大学院工学研究院 電力・エネルギー工学
餘利野 直人，造賀 芳文，佐々木 豊