Deep Learning

Research and Applications

Siddhartha Bhattacharyya, Vaclav Snasel, Aboul Ella Hassanien, Satadal Saha, B. K. Tripathy, Siddhartha Bhattacharyya, Vaclav Snasel, Aboul Ella Hassanien, Satadal Saha, B. K. Tripathy

• Editorial: De Gruyter
• Año: 2020
• ISBN:
  9783110670929
• Edición: 1
• Categoría: Computer Science
• Categoría: Artificial Intelligence (AI) & Semantics

This book focuses on the fundamentals of deep learning along with reporting on the current state-of-art research on deep learning. In addition, it provides an insight of deep neural networks in action with illustrative coding examples.

Deep learning is a new area of machine learning research which has been introduced with the objective of moving ML closer to one of its original goals, i.e. artificial intelligence. Deep learning was developed as an ML approach to deal with complex input-output mappings. While traditional methods successfully solve problems where final value is a simple function of input data, deep learning techniques are able to capture composite relations between non-immediately related fields, for example between air pressure recordings and English words, millions of pixels and textual description, brand-related news and future stock prices and almost all real world problems.

Deep learning is a class of nature inspired machine learning algorithms that uses a cascade of multiple layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. The learning may be supervised (e.g. classification) and/or unsupervised (e.g. pattern analysis) manners. These algorithms learn multiple levels of representations that correspond to different levels of abstraction by resorting to some form of gradient descent for training via backpropagation. Layers that have been used in deep learning include hidden layers of an artificial neural network and sets of propositional formulas. They may also include latent variables organized layer-wise in deep generative models such as the nodes in deep belief networks and deep boltzmann machines. Deep learning is part of state-of-the-art systems in various disciplines, particularly computer vision, automatic speech recognition (ASR) and human action recognition.

1 Deep Learning – A State-of-the-Art Approach to Artificial Intelligence

Soumyajit Goswami

IBM India Private Limited, Salt Lake, Sector V, Kolkata, West Bengal, India

Abstract

This chapter presents various cloud platforms that are available in market offerings from different vendors. IBM provided a machine learning (ML) platform “IBM Watson Studio” (formerly “Data Science Experience”), and this is considered here for the field of study. An overview of artificial intelligence, ML, and deep learning (DL) with their relationship is deliberated. Discussion on popular DL architectures with elementary comparison is also considered.

Keywords: Artificial intelligence, machine learning, deep learning, IBM Watson Studio (formerly, Data Science Experience or DSX),

1.1 Introduction

Deep learning (DL), the subfield of artificial intelligence (AI), is the most promising area considered for research and industry. Although DL is a very modern topic, it is already being used by multiple technology giants to fulfill their needs. Few examples are voice and image recognition algorithms of Google [1]: Netflix and Amazon use it to decide [2] which video a person desires to watch or purchase in near future, upcoming forecast by MIT researchers [3], and Facebook uses it to predict future actions for advertisers [4]. UCLA researchers have manufactured an advanced microscope that produces a high-dimensional dataset used to train a DL network in identifying cancer cells in tissue samples [5]. In a nutshell, it has been used nowadays everywhere whenever automation comes into picture.

In the following section of this chapter, the relationship between DL, machine learning (ML), and AI has been discussed. A brief introduction to artificial neural network (ANN) with its classification and its different learning techniques has been specified in Section 1.3. As part of classification of ANN, feedforward neural networks (FFNNs) and recurrent neural networks (RNNs) with their uses have been discussed. While in the section of learning techniques, supervised, unsupervised, and reinforcement learning are briefly considered. Section 1.4 has been reserved to discuss about DL. It has been stated clearly in this section why the term “deep” has been used. Multiple points have been identified, which makes DL as state of the art. In Section 1.6, different activation functions such as sigmoid activation function, hyperbolic tangent activation function, rectified linear unit (ReLU) activation function, and softmax activation function are described in detail. There are many DL architectures available in literature. Few of them became very popular and offers high accuracy resulting in better performance. The concepts of restricted Boltzmann machine (RBM), deep belief network (DBN), autoencoder (AE), and convolutional neural network (CNN) are deliberated in this section. In Section 1.6, multiple ML platforms from different organizations have been furnished. All of them provide cloud infrastructures with high-performance graphics processing units (GPUs) to quicken the training of DL network with the huge volumes of data, which lessens the training time from weeks to hours. The last section of this chapter is dedicated for describing different steps of using IBM ML platform – IBM Watson Studio (formerly, Data Science Experience or DSX).

1.2 AI versus ML versus DL

AI is a subcategory of computer science that handles the simulation of intelligent activities in computers. AI is a computer system, which can accomplish responsibilities that usually need human acumen. Generally, a rule engine leads the AI system and a good AI system should have an intelligent rule engine, which is based on a series of meaningful IF–THEN statements. Since the 1950s, AI has been successfully used in visual perception, speech recognition, decision-making, and translation between languages. AI and ML are often used interchangeably, especially in the realm of big data.

As shown in Figure 1.1, DL is considered as a subcategory of ML and again ML is a subcategory of AI. In other words, all DL i...

Índice

• Title Page
• Copyright
• Contents
• Dedication
• Preface
• List of Contributors
• 1 Deep Learning – A State-of-the-Art Approach to Artificial Intelligence
• 2 Convolutional Neural Networks: A Bottom-Up Approach
• 3 Handwritten Digit Recognition Using Convolutional Neural Networks
• 4 Impact of Deep Neural Learning on Artificial Intelligence Research
• 5 Extraction of Common Feature of Dysgraphia Patients by Handwriting Analysis Using Variational Autoencoder
• 6 Deep Learning for Audio Signal Classification
• 7 Backpropagation Through Time Algorithm in Temperature Prediction
• Index