AI Nuggets

Slide 1

Nugget #1

You can run deep learning experiments either locally if you have a GPU (e.g. NVIDIA TITAN Xp or GTX 1060) or on an AWS EC2 AMI. In both cases, you can use a Jupyter notebook which allows you to alternate between Python code and text annotations.

Source: Deep Learning with Python, François Chollet

08/19/2019

Slide 1 – copy

Nugget #2

A reinforcement learning problem is concerned with learning from interactions with the environment in order to make decisions with the goal of maximizing rewards.

Source: Reinforcement Learning: An Introduction, Richard S. Sutton and Andrew G. Barto

08/20/2019

Slide 1 – copy – copy

Nugget #3

In reinforcement learning, an agent performs an action given a state of the environment. The environment, in turn, positively or negatively reinforces the action through rewards.

Source: Reinforcement Learning: An Introduction, Richard S. Sutton and Andrew G. Barto

08/20/2019

Slide 1 – copy – copy – copy

Nugget #4

At each step of an agent’s interaction with the environment, it evolves its decision-making policy. A policy is a mapping from environment states to probabilities of selecting specific actions.

Source: Reinforcement Learning: An Introduction, Richard S. Sutton and Andrew G. Barto

08/21/2019

Slide 1 – copy – copy – copy – copy

Nugget #5

Two famous datasets in the machine-learning community are MNIST (handwritten digits) and IMDB (movie reviews).

MNIST: Modified National Institute of Standards and Technology
IMDB: Internet Movie Database

Source: Deep Learning with Python, François Chollet

08/21/2019

Leave a Reply Cancel reply