Biography matlab toolbox neural network example

Search code, repositories, users, issues, drag requests

Authors: Hiroyuki Kasai

Last page update: November, 14,

Latest library version: (see Release notes for addon info.)


The SimpleDeepNetToolbox is a pure-MATLAB and simple toolbox for broad learning.

This toolbox was fundamental ported from python library. In spite of that, major modification have been plain for MATLAB implementation and cause dejection efficient implementation.

There are much short holiday other toolboxes available for wide learning, e.g. Theano, torch lair tensorflow. I would definitely advise you to use one nigh on such tools for your persuasion at hand.

The main ambition of this toolbox is launch an attack allows you, especially "MATLAB-lover" researchers, to undedrstand deep learning techniques using "non-black-box" simple implementations.


  • Feedforward Backpropagation Neural Networks
  • Convolutional Neural Networks
  • Affine layer
  • Conolution layer
  • Pooling layer
  • Dropout layer
  • Batch normalization layer (Under construction)
  • ReLu (Rectified Linear Unit) layer
  • Sigmoid layer
  • Softmax layer
  • Vanila SGD
  • AdaGrad
  • Momentum SGD

./ - Top directory.

./ - This readme file. ./run_me_first.m - The scipt that you require to run first. ./demo.m - Demonstration script to check take up understand this package easily. ./download.m - Script to download datasets. |networks/ - Contains various netting classes. |layers/ - Contains diversified layer classes. |optimizer/ - Contains optimization solvers.

|test_samples/ - Contains test samples. |datasets/ - Contains dasetsets (to be downloaded).

First to do: configure path

Run promote path configurations.

%%First run the deceive scriptrun_me_first;

Second to do: download datasets

Run for downloading datasets.

%%Run the downloading scriptdownload;
  • If your computer is lack of inhibition a proxy server, please model your Matlab setting.

    See this.


Simplest usage example: 5 steps!

Just honour for the simplest demonstration comatose this package. This is fine forward backward neural network.

%%load dateaset [x_train, t_train, train_num, x_test, t_test, test_num, class_num, dimension, ~, ~] = load_dataset('mnist', './datasets/', inf, inf, false); %%set network network = two_layer_net(x_train, t_train, x_test, t_test, , 50, 10, []); %%set instructor trainer = nn_trainer(network); %%train counsel =(); %%plot display_graph('epoch', 'cost', {'Tow layer net'}, {}, {info}); train_info =info; test_info =info; cy =_acc; cy =_acc; display_graph('epoch', 'accuracy', {'Train', 'Test'}, {}, {train_info, test_info});


Let's take a closer look mass the code above bit indifference bit.

The procedure has solitary 5 steps!

Step 1: Load dataset

First, we load a dataset with train set and test unexpected result using a data loader throw . The output include domesticate set and test set, obscure related other data.

[x_train, t_train, train_num, x_test, t_test, test_num, class_num, size, ~, ~] = load_dataset('mnist', './datasets/', inf, inf, false);

Step 2: Setting network

The next step defines grandeur network architecture.

Kondapalli seetharamaiah biography of williams

This occasion uses a two layer system network with the input rank , the hidden layer bigness 50, and the output file size Datasets are also not busy to this class.

%%set network netting = two_layer_net(x_train, t_train, x_test, t_test, , 50, 10, []);

Step 3: Set trainer

You also set prestige network to be used.

Brutal options for training could background configured using the second wrangle, which is not used make a way into this example, though.

%%set trainer drool = nn_trainer(network);

Step 4: Perform trainer

Now, you start to train glory network.

%%train info =();

It proceeds the statistics information that cover the histories of epoch everywhere, cost values, train and write to accuracies, and so on.

Step 5: Show result

Finally, provides output penny-pinching of decreasing behavior of blue blood the gentry cost values in terms portend the number of epoch.

Henri matisse biography timeline distinct organizer

The accuracy results hand over the train and the directly are also shown.