I am happy to announce that my second research publication got accepted for presentation during the International Conference on Neural Information Processing (ICONIP) This year it is taking place in Auckland, New Zeland.
The paper is entitled “Efficient Pruning and Compression Techniques for Convolutional Neural Networks to Preserve Knowledge and Optimize Performance”. I have coauthored it with Professor Adrian Horzyk [ORCID: 0000−0001−9001−4198]. It concerns very important topic of optimisation of high performance CNNs (Convolutional Neural Networks). The paper introduces novel algorithm of pruning and merging convolutional filters allowing for over 86% FLOPs rediction while preserving the original performance of the network. Key advantage of our method is that it requires much less time than our competitiors.