The IBS-Yonsei research team introduces a novel Lp-Convolution method at ICLR 2025. A team of researchers from the (IBS), ...

Despite the widespread success of neural networks, their susceptibility to adversarial examples remains a significant challenge. Adversarial training (AT) has emerged as an effective approach to ...

The effectiveness of this approach has led to the development of various neural network architectures, including convolutional neural networks (commonly used in image recognition) and transformer ...

In this paper, an optical convolutional neural network, combining a novel architectural design with a compatible data encoding technique, is introduced to achieve superior data efficiency by ...

Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization; (iii) Structuring Machine Learning Projects; (iv) Convolutional Neural Network… ...

Additionally, our method enables data embedding in different layers of neural networks, including linear layers, convolutional layers, and transpose convolutional layers. In cover networks, the hidden ...

Methods: We developed a computational MDA prediction method called GPUDMDA by combining graph attention autoencoder, positive-unlabeled learning, and deep neural network. First, GPUDMDA computes ...

Released in 2024, Flax NNX is a new simplified Flax API that is designed to make it easier to create, inspect, debug, and analyze neural networks in JAX. It achieves this by adding first class support ...

School of Management, Dalian Polytechnic University, Dalian 116034, P. R. China School of Information Science and Engineering, Dalian Polytechnic University, Dalian 116034, P. R. China ...

The network comprises 22 layers that require training (or 27 if pooling layers). Experiments have shown that GoogLeNet has fewer trainable weights than AlexNet and, thus, is more accurate (Szegedy et ...