====== Products ======
==== Software Products ====
==== Deep blue ====
IBM, chess\\

==== AlphaGo ====
Google Deep Mind\\
plays the ancient Chinese game of Go\\
consistently defeats the world's best human Go players\\
Alpha go uses two neural networks\\
  - the policy net\\
2. the value net\\
A new upgrade AlphaGoZero, defeats the old AlphaGo soundly\\

==== Services  ====
==== Google Cloud Platform (GCP) ====
offering paid services by the minute\\
Google Cloud Organization\\
==== Universe, OpenAI ====

but if you want to rent your own infrastructure you will typically be charged by the minute (Google) or hour (Amazon, Microsoft).

==== Simulators ====

3D Learning Enviornment.  For training AI agents.

== Unity Machine Learning Agents ==
Software to turn games made via the Unity game engine into environments in which to train AI systems.

== UE4 ==
//Unreal 4// game engine used as simulation for vision training within Unity

==== APIs, Tools, Frameworks ====
==== Tensorflow ====
Express a numeric computation as a graph.\\
Data flow graph with nodes (aka vertices or points) and edges (aka lines or arcs).\\
Graph nodes are operations. (This is theoretical.  Sometimes a node represents a variable.)\\
Graph edges are tensors which flow between nodes.  (per Stanford lecture)\\

Types of graph nodes:\\
  * Variable
  * Placeholder
  * Mathematical operations

Tensorboard - a second product to visualize the graph\\

Internet opinion: TensorFlow is a //huge// success for Google, having quickly become the default AI programming framework used by newbies, Googlers, and established teams outside of Google, to train and develop AI systems, and also creating a larger and larger set of software developers primed to use its many cloud services, which tend to require or gain additional performance boosts from using TensorFlow.

==== SciKit Learn ====
Does logistic regression, k nearest neighbor, support vector machines.\\
Does Not do neural networks. Does NOT have gpu support.\\
(Use anaconda to install with dependencies.)\\

==== Caffe ====
The first, 2013, but now passe.

==== Theano ====
Developed and managed by MILA.\\
Discontinued as of 28 Sep 2017, per Yoshua Bengio, Director MILA  https://groups.google.com/forum/#!topic/theano-users/7Poq8BZutbY\\

==== Keras ====
high-level, sits on top of tensorflow and theano\\

==== tflearn ====
high-level, sits on top of tensorflow\\
Opinion on the internet: "Use keras on top of theano, or, use tflearn on top of tensorflow.\\

==== Universe ====
product released by openai, uses tensorflow

==== Google Cloud API’s ====
  * Machine Learning Engine
  * Jobs
  * Natural Language
  * Speech
  * Translation
  * Vision
  * Video Intelligence

==== Kaggle ====
acquired by google cloud\\
builds datasets\\

==== Robotics Operating System (ROS) ====
http://www.ros.org
Open source api and framework to control robots


==== Proprietary Frameworks ====

  * Amazon offers MXNET\\
  * Apple offers CoreML, for MacOS, iOS, watchOS, and tvOS.
  * Microsoft offers CNTK\\
  * Facebook offers PyTorch and Caffe2

In an attempt to compete with TensorFlow,
Facebook and Microsoft announced the Open Neural Network Exchange (ONNX) format which support their products, and Amazon and Apple have announced they are working together toward interoperability between their products.