#deep learning
Deep Learning: 18 AI articles covering deep learning news, analysis, and research
Articles
Architectural vs. Regularization Enhancements in Deep Learningโญ9
Ablation study shows regularization losses drive 21% accuracy gains and 99% fewer deformations in deep learning MRI registration, with affine stages adding mode...
FVeinSyn: Synthetic Finger Vein Image Generatorโญ8
FVeinSyn generates 500,000 synthetic finger vein images, boosting recognition accuracy by 27.43% across eight public datasets.
Deep Learning Automates Parameter Extraction for 2D Transistors:โญ9
Deep learning automates 2D transistor parameter extraction, boosting speed and consistency for nanoscale semiconductor characterization and modeling.
The Sigmoid Function: From 'e' to Neural Networksโญ9
Learn how the sigmoid function maps inputs to probabilities, its derivation from odds, and its role in neural networks.
Neural Networks Explained: What They Are and How to Build One inโญ10
Learn what neural networks are and how to build one from scratch in Python using NumPy, explained simply for beginners.
Backpropagation Explained for Beginners (Part 3): Howโญ9
Learn how backpropagation really works in neural networks, explained simply with chain rule gradients, shared derivatives, and step-by-step examples for beginne...
Variational Autoencoders (VAEs) Explained: From Theory to ELBOโญ9
Discover how Variational Autoencoders (VAEs) work, from theory to ELBO and the reparameterization trick, explained clearly for machine learning enthusiasts.
SPP-Net Paper Walkthrough: Breaking the Fixed-Size Constraintโญ10
Learn how SPP-Net removes the fixed-size input constraint in CNNs, enabling variable image sizes without cropping or warping, with a PyTorch implementation guid...
IMDb Sentiment Analysis with DistilBERT LoRA, TF-IDF Baselines,โญ9
Explore IMDb sentiment analysis with DistilBERT LoRA, TF-IDF baselines, calibration, robustness tests, and semi-supervised learning.
Advancing Utility Pole and Sign Detection Through Deep Learningโญ10
Deep learning detects utility poles and signs from street imagery, estimating lean angles with 90% mAP via modified DETR, outperforming standard detectors.