Results 111 to 120 of about 118,488 (274)

CSD: Channel Selection Dropout for Regularization of Convolutional Neural Networks

IEEE Access
In this study, we present a novel approach, Channel Selection Dropout (CSD), designed to regularize deep convolutional neural network (CNN) architectures. Unlike standard Dropout, which randomly deactivates neurons in fully connected layers, CSD works on
Imrus Salehin, Dae-Ki Kang
doaj   +1 more source

Compacting Neural Network Classifiers via Dropout Training

CoRR, 2016
Submitted to AISTATS 2017 (Short-version is accepted to NIPS Workshop on Efficient Methods for Deep Neural Networks)
Yotaro Kubo, George Tucker, Simon Wiesler   +2 more
openaire   +2 more sources

Application of Neural Networks for Advanced Ir Spectroscopy Characterization of Ceria Catalysts Surfaces

Advanced Intelligent Discovery, EarlyView.
A novel convolutional neural network architecture enables rapid, unsupervised analysis of IR spectroscopic data from DRIFTS and IRRAS. By combining synthetic data generation with parallel convolutional layers and advanced regularization, the model accurately resolves spectral features of adsorbed CO, offering real‐time insights into ceria surface ...
Mehrdad Jalali, Lachlan Caulfield, Eric Sauter, Alexei Nefedov, Chengwu Yang, Christof Wöll   +5 more
wiley   +1 more source

Accelerating Primary Screening of USP8 Inhibitors from Drug Repurposing Databases with Tree‐Based Machine Learning

Advanced Intelligent Discovery, EarlyView.
This study introduces a tree‐based machine learning approach to accelerate USP8 inhibitor discovery. The best‐performing model identified 100 high‐confidence repurposable compounds, half already approved or in clinical trials, and uncovered novel scaffolds not previously studied. These findings offer a solid foundation for rapid experimental follow‐up,
Yik Kwong Ng, Sharon Ai Er Lam, Wei Chun Lim, Hannah Hui Min Toh, Shi Jun Ang   +4 more
wiley   +1 more source

Multimodal machine learning framework for predicting and enhancing higher education using reinforcement learning and graph neural networks

Automatika
There has been a growing emphasis on intelligent educational systems to foster student success, with universities working to aid timely interventions and personalized learning.
Chen Jie, Huang Min, Chen Bin
doaj   +1 more source

Parallel Dither and Dropout for Regularising Deep Neural Networks

CoRR, 2015
Effective regularisation during training can mean the difference between success and failure for deep neural networks. Recently, dither has been suggested as alternative to dropout for regularisation during batch-averaged stochastic gradient descent (SGD).
openaire   +2 more sources

Data‐Driven High‐Throughput Volume Fraction Estimation From X‐Ray Diffraction Patterns

Advanced Intelligent Discovery, EarlyView.
Long exposure times and the need for manual evaluation limit the use of X‐ray diffraction in high‐throughput applications. This study presents a data‐driven approach addressing both issues. HiVE (a method for High‐throughput Volume fraction Estimation) performs composition estimation for high‐noise XRD patterns produced using polychromatic emission ...
Hawo H. Höfer, André Orth, Robert Wang, Ben Breitung, Simon Schweidler, Jasmin Aghassi‐Hagmann, Markus Reischl   +6 more
wiley   +1 more source

Harnessing Machine Learning to Understand and Design Disordered Solids

Advanced Intelligent Discovery, EarlyView.
This review maps the dynamic evolution of machine learning in disordered solids, from structural representations to generative modeling. It explores how deep learning and model explainability transform property prediction into profound physical insight.
Muchen Wang, Yue Fan
wiley   +1 more source

Why Physics Still Matters: Improving Machine Learning Prediction of Material Properties With Phonon‐Informed Datasets

Advanced Intelligent Discovery, EarlyView.
Phonons‐informed machine‐learning predictive models are propitious for reproducing thermal effects in computational materials science studies. Machine learning (ML) methods have become powerful tools for predicting material properties with near first‐principles accuracy and vastly reduced computational cost.
Pol Benítez, Cibrán López, Edgardo Saucedo, Teruyasu Mizoguchi, Claudio Cazorla   +4 more
wiley   +1 more source

Parametric Analysis of Spiking Neurons in 16 nm Fin Field‐Effect Transistor Technology

Advanced Intelligent Discovery, EarlyView.
Energy efficient computing has driven a shift toward brain‐inspired neuromorphic hardware. This study explores the design of three distinct silicon neuron topologies implemented in 16 nm fin field‐Effect transistor technology. While the Axon‐Hillock design achieves gigahertz throughput, its functional fragility persists. The Morris–Lecar model captures
Logan Larsh, Raiyan Siddique, Sarah S. Sharif, Yaser Mike Banad   +3 more
wiley   +1 more source