tensorflow - aalto · tensorflow api tensorflow has apis available in several languages both for...

Tensorflow

Does everything in one click!

Does everything in one click!

Machine learning (Neural nets, classification, regression, clusterization, cross-validation, batch-learning, gradient boosting, convex optimization etc.)

Does everything in one click!

Machine learning (Neural nets, classification, regression, clusterization, cross-validation, batch-learning, gradient boosting, convex optimization etc.)

Few lines of code! (High level intuitive python API)

Does Machine learning in few lines of code!

What is Tensorflow?

An open-source software library for Machine Intelligence

Why is Tensorflow different?

TensorFlow™ is an open source software library for numerical computation using data flow graphs. Nodes in the graph represent mathematical operations, while the graph edges represent the multidimensional data arrays (tensors) communicated between them.

Companies using TensorFlow

Tensorflow API

TensorFlow has APIs available in several languages both for constructing and executing a TensorFlow graph. The Python API is at present the most complete and the easiest to use, but other language APIs may be easier to integrate into projects and may offer some performance advantages in graph execution.

Python

C++

Java

Go

TensorFlow API

TensorFlow provides multiple APIs.

The lowest level API - TensorFlow Core - provides you with complete programming control. We recommend TensorFlow Core for machine learning researchers and others who require fine levels of control over their models.

A high-level API like tf.contrib.learn helps you manage data sets, estimators, training and inference.

Why data flow graphs at all?

The flexible architecture allows you to deploy computation to one or more CPUs or GPUs in a desktop, server, or mobile device with a single API.

Tensorflow prerequisites

• How to program in Python

• At least a little bit of knowledge about arrays

• Ideally, something about machine learning

Tensorflow data flow

Core graph data structures tf.Graph

tf.Operation

tf.Tensor

tf.Graph

tf.Operation is a node, represents an operation.

Accepts a tf.Tensor and returns a tf.Tensor.

tf.Tensor is an edge, represents a piece of data of a specific array shape (scalar or 1D-array or matrix etc.).

Standard TensorFlow libraries

Math, Strings, Histograms, Control flow, Higher Order functions, TensorArray operations, Tensor Handle operations, Images, Sparse Tensors, Inputs and Readers, Data IO, Neural Networks, RNN, Training, Summary operations, Testing, BayesFlow, CRF, Statistical distributions, Random variable transformations, FFmpeg, Layers, Learn, Monitors, Losses, Metrics.

Small code example (convex problem)

Tensor board TensorBoard is a suite of web applications for inspecting and understanding your

TensorFlow runs and graphs.

It can output the graph of the calculation and draw a chart of summary values which

you log in the code.

MNIST DEMO

MNIST Linear Softmax model

MNIST DEMO graph

MNIST Convolutional NN example

Conclusion

•High level and powerful machine learning library for Python with single API for CPU and GPU

•Requires a bit of getting familiar with the syntax, then you can easily create powerful and scalable machine learning models