#deep learning

Notes on the Transformer architecture: attention mechanisms, multi-head attention, positional encoding, and encoder/decoder layers with PyTorch implementations.

A compilation of interesting papers I enjoyed reading, including I-Con, Denoising Diffusion Probabilistic Models, and SpiRobs.

Notes on the mathematical foundations of diffusion models, including the noising process, denoising process, loss function, and training/generation loops.

A talk covering miniaturization techniques for deep learning models operating in real-time hearing aid environments.

A study of Acoustic Scene Classification (ASC) for hearing aids, leading to the development of an industrial ASC model based on a convolutional neural network designed to run in hearing aids for noise suppression purposes.

Implementation of the Denoising Diffusion Models paper on CIFAR-10, using a pretrained U-Net and sinusoidal time embeddings.

A lightweight and flexible early stopping utility for PyTorch training loops, inspired by the Keras/TensorFlow EarlyStopping callback.

Implementation of transformer encoder-decoder model based on the Attention is all you need paper.

Miniaturizing Acoustic Scene Classification from 864 MFLOPs to 1.6 MFLOPs using Knowledge Distillation and Depthwise Convolutions.

An introductory course on neural networks applied to digital signal processing and sound synthesis.