Quoc Le's Lectures on Deep Learning — Gaurav Trivedi
deep-learningneural-networkslecturesgoogle-brainmlss-2014
Abstraction: Content guide for Quoc Le's 3-lecture deep learning series at MLSS 2014
Key points:
- Three 4K video lectures delivered on a blackboard at Machine Learning Summer School (MLSS '14) in Pittsburgh; no slides provided
- Lecture 1 covers neural network basics: decision functions, loss minimization, gradient descent, backpropagation derivation
- Lecture 2 covers practical NN training: SGD, breaking symmetry, underfitting/overfitting, hyperparameter selection, ReLU, autoencoders for initialization
- Lecture 3 covers architectures: ConvNets (local receptive fields, image classification pipeline), Recurrent NNs, word vectors, Google Brain applications
- Quoc Le is from the Google Brain team that made headlines in 2012 for the unsupervised cat-recognizer trained on YouTube frames
- Full video series available on YouTube (4 videos) from mlss2014.com materials page
Connections: Quoc Le · Google Brain · Deep Learning · Neural Networks · Convolutional Neural Networks