ANN Implementation

Author : Zhekai Jin (Scott)

Description

An implementation of Artificial Neural Network using C++.

Dependency

• CMAKE 3.0
• MAKE

Usage

./ANN_train
Follow the Prompt
./ANN_test
Follow the Prompt

Build

https://github.com/ZhekaiJin/Artifical_NN.git
cd Artifical_NN/
mkdir build && cd build
cmake .. && make 

Clear Build

cd Artifical_NN/
rm -rf build

Database Description

• Filename: robot_dataset

• Title: Wall-Following navigation task with mobile robot SCITOS-G5

• Reasonable parameter:

  1. learning rate : 0.05
  2. Epoch(iterations) : 100
  3. Number of hidden nodes: 5

• Filename for trained/untrained/result:

  1. untrained: init_weight
  2. trained: sample.out.05.100.trained
  3. result: sample.out.05.100.results

• Initial weights are generated through scirpt : init_weight.py

• Database are modified using script : format_to_data.py [take 1 parameter as the percentage of the training set to be partitioned from the dataset, e.g. 0.8]

  1. Dropped incomplete data
  2. Encoded the classification to binary classification of four nodes
  3. Drop out irrelevant feature (no correlation to classification task)
  4. Partition the data to test (20 % of original sets) and train (80% of original sets)

• Full description of the dataset can be found at Wall-following.names in the folder, dataset found on UCI machine learning dataset repository