- Backprop by Hand and CNNs on MNIST
Hand-computing one step of backpropagation, then iterating on CNN architectures in MATLAB until giving up and switching to Keras.
5 min - Fractals for Image Classification Feature Extraction
Exploring how fractal geometry, specifically Hilbert curves, can provide a novel approach to feature extraction in image classification.
11 min - BiLSTM-CNN-CRF for Sequence Labeling: A Tutorial
A PyTorch tutorial for the ACL 2016 paper on End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF, presented at ICML 2018.
13 min - Deep Learning Methods for Quotable Text
Exploring what makes a sentence memorable using deep learning, achieving significant improvements over traditional linguistic feature approaches.
9 min
Back