ANN_ParallelComputing

Overview

ANN_ParallelComputing is a deep learning project aimed at exploring a simple Artificial Neural Network (ANN) architecture and implementing parallel computing through OpenMP and OpenCL to enhance the performance of training and inference. This project focuses on deep learning applications using the MNIST dataset and CIFAR-10 dataset.

Features and Goals

• Use a simple ANN architecture for deep learning.
• Implement parallelization on multi-core CPUs using OpenMP to accelerate the training process.
• Implement parallelization on GPUs or other accelerators using OpenCL for further performance improvement.
• Provide training and testing functionality for both MNIST and CIFAR-10 datasets.
• Support custom network configurations and hyperparameter tuning.

Key Components

mnist.hpp

Contains functionalities and structure definitions related to the MNIST dataset. The MNIST dataset is a classic handwritten digit recognition dataset commonly used for testing and experiments in deep learning and machine learning.

readCIFAR10.hpp

Contains the definition of the readCIFAR10 function, which is used to read the CIFAR-10 dataset. CIFAR-10 is a widely used image classification benchmark dataset containing images from multiple classes.

initGPU.hpp

Contains functionalities for initializing GPU resources and uses OpenCL as the library for GPU computations. This library aids in configuring and preparing the GPU for parallel computation.

MNIST.hpp

Contains functionalities and structure definitions for processing image data from the MNIST dataset. The MNIST dataset is a classic handwritten digit recognition benchmark dataset commonly used for experiments in deep learning and machine learning.

ANN.cpp

Contains the implementation of a program for training an Artificial Neural Network (ANN). This program is used for building, training, and evaluating ANN models, utilizing the backpropagation algorithm for training.

ANNOCL.cpp

An implementation of ANN.cpp with OpenCL integration.

ANNOMP.cpp

An implementation of ANN.cpp with OpenMP integration.

CIFAR10.cpp

Contains a C++ function for reading the CIFAR-10 dataset.

GPU.cpp

Contains functionalities for executing GPU computations using OpenCL.

main.cpp

Contains an example of implementing a deep learning autoencoder using the code.

MNIST.cpp

Contains a C++ function for reading the MNIST dataset.

Acknowledgments

This project is inspired by and receives support from Professor You-Ming Hsieh, and we appreciate the files and guidance provided by him. This project is for learning and educational purposes only and is not intended for any commercial use.