ARTIFICIAL NEURAL NETWORK

Ms. Sarita Assistant Professor from Computer Science & Engineering Department of   Dronacharya College of Engineering, Gurgaon, CSE Department attended a Session on “Advances in pattern classification using Artificial Neural Network, Spiking Neural Network and Decision Tree” at MZ 168, Committee Room, Department of Mathematics, IIT Delhi On 6th June 2014.

In this Session Mr. Venkata Naresh babu kuppili, Research Scholar at IIT Delhi presents an algorithm that constructs feed-forward neural networks with a single hidden layer for pattern classification. The algorithm starts with a small number of hidden units in the network and adds more hidden units as needed to improve the network's predictive accuracy. To determine when to stop adding new hidden units, the algorithm makes use of a subset of the available training samples for cross validation. New hidden units are added to the network only if they improve the classification accuracy of the network on the training samples and on the cross-validation samples. Extensive experimental results show that the algorithm is effective in obtaining networks with predictive accuracy rates that are better than those obtained by state-of-the-art decision tree methods.

Attendees during this presentation were Dr. B. S. Panda (HOD, Department of Mathematics, IIT-D), Prof B. Chandra (IIT-D), Prof Aparna Mehra (IIT-D), Prof Niladri Chatterjee (IIT-D) & Prof Amitabha Tripathi (IIT-D).

Speaker also proposed an improved architecture for Probabilistic Neural Networks (IAPNN) with an aggregation function based on f-mean of training patterns. The improved architecture has reduced number of layers and that reduces the computational complexity. It is observed from the performance evaluation on various benchmark datasets that IAPNN outperforms in terms of classification accuracy. A biologically realistic spiking neuron model has been also proposed by speaker which contains a novel non linear spiking function. Proposed neuron model contains a lower order spike generating function in contrast to the spike generating function of Quadratic integrate fire neuron model.

Overall Session was very informatics and gave knowledge of latest Research in the area of Neural Network.