Goodfellow, Bengio, Courville - «Deep Learning» (2016)

dmitrii_fediuk (Dmitrii Fediuk) May 16, 2019, 11:19pm 1

What is a «hidden layer» of a neural network?

What is a «feature»?

Increasing model sizes

What is a «neural Turing machine»?

Limitations of linear models

What are «factors of variation»?

What is «reinforcement learning»?

What has neuroscience given to deep learning?

What is the «computational neuroscience»?

What is «autoencoder»?

What is «connectionism»?

What is «distributed representation»?

Deep neural networks in the mid-1990s - mid-2000s

The third wave of neural networks research

Increasing dataset sizes

What is «representation learning»?

What is «Big Data»?

Simple machine learning algorithms

How many training examples is needed for a supervised learning?

The number of neurons in animals and artificial neural networks

Сhoice of representation

Increasing accuracy, complexity and real-world impact

2 main ways of measuring the depth of a deep learning model

What is a «layer» of a neural network?

What is «deep learning»?

Historical trends in deep learning

What is a «visible layer» of a neural network?

The deep learning's history

Why has the neuroscience's role in deep learning been diminished?

What is «disentangling factors of variation»?

What is «machine learning»?

Simple linear models: predecessors of modern deep learning