Abstract

The main purpose of this project is to modify a convolutional neural network for image classification, based on a deep-learning framework. A transfer learning technique is used by the MATLAB interface to Alex-Net to train and modify the parameters in the last two fully connected layers of Alex-Net with a new dataset to perform classifications of thousands of images. First, the general common architecture of most neural networks and their benefits are presented. The mathematical models and the role of each part in the neural network are explained in detail. Second, different neural networks are studied in terms of architecture, application, and the working method to highlight the strengths and weaknesses of each of neural network. The final part conducts a detailed study on one of the most powerful deep-learning networks in image classification – i.e. the convolutional neural network – and how it can be modified to suit different classification tasks by using transfer learning technique in MATLAB.

Highlights

  • 1.1 The Success of NNs and the Type of Problems NNs AddressArtificial neural networks (NN) are computational models derived from the simulation of typical human brain activities such as image classification, pattern recognition, and language understanding

  • 4.3 Training convolutional neural network (CNN) The training phase of a CNN is much harder than the typical ordinary neural network (ONN), as the input color image that has 200 by 300 pixels is fed to a CNN as 200 by 300 by 3 pixels

  • The key objective of this project was to design a neural network that can perform image classification based on a convolutional neural network (CNN)

Read more

Summary

The Success of NNs and the Type of Problems NNs Address

Artificial neural networks (NN) are computational models derived from the simulation of typical human brain activities such as image classification, pattern recognition, and language understanding. Computer programs are very useful tools for performing tasks with high speed, accuracy, and reliability. These information processing systems are not intelligent, which requires human intervention for updating. Critical-based neural networks utilize a supervisor to evaluate the performance of the network, based on as specific cost function This type of network is used for robust optimal tracking control [6] and data-driven control systems [7]. A convolutional neural network (CNN) is a major type of DNN, which can be applied for classification, regression, and image recognition. The way CNNs can classify images mainly depends on the architecture of the neural network (e.g. the type, size, and order of the layers). The layer itself comes with many parameters such as weights and bias that play a significant role in the network performance

The Architecture of Neural Network
Fundamental Building Block and Parameters
Multilayer NNs and Activating function
Deep Neural Networks
Objective of the Thesis
Chapter 2 Ordinary Neural Networks
Single Layer Neural Networks
Convex Optimization
Time Delayed Versions of the inputs
Sliding Window Technique
Back Propagation Through Time
Why ONNs are Not enough
Rectified Linear Unit (ReLU)
Training CNNs
CNNs and Graphic Processing Unit (GPUs)
Transfer Learning
Chapter 5 Numerical Result
Classifying Multiple Sub-Categories of Mammals
Findings
Conclusion
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.