Identity management using biometrics--a fascinating research area--adds another dimension when coupled with deep learning techniques. Deep learning for biometrics, organized into four parts, highlights new possibilities within the domain.

Part 1, “Deep Learning for Face Biometrics,” consists of three chapters. Chapter 1, “The Functional Neuroanatomy of Face Processing,” attempts to correlate functions of different regions of the human brain to different layers in convolutional neural networks (CNNs), and recommends a deep CNN like FaceNet for modeling purposes. Chapter 2, “Real-Time Face Identification via Multi-convolutional Neural Network and Boosted Hashing Forest,” uses a two-stage learning approach--learning human faces using a CNN, and then transforming it by appending a boosted hashing forest technique--yielding improved results. Chapter 3, “CMS-RCNN: Contextual Multi-Scale Region-Based CNN for Unconstrained Face Detection,” attempts to solve the problem of face detection under adverse conditions like feature occlusion and improper illumination.

Part 2, “Deep Learning for Fingerprint, Fingervein and Iris Recognition,” also contains three chapters. In chapter 4, “Latent Fingerprint Image Segmentation Using Deep Neural Network,” image patches of fingerprints are used to train a deep artificial neural network (DANN) and later to recreate original latent fingerprints. Chapter 5 is “Finger Vein Identification Using Convolutional Neural Network and Supervised Discrete Hashing.” Vascular biometrics located a little beneath the skin’s surface are used to train different CNN architectures. The modified VGG-16 model achieves higher accuracy than other models. Chapter 6, “Iris Segmentation Using Fully Convolutional Encoder-Decoder Networks,” uses three types of fully convolutional encoder-decoder network (FCEDN) architectures for segmenting iris images.

Part 3, “Deep Learning for Soft Biometrics,” contains four chapters. Chapter 7, “Two Stream CNNs for Gesture-Based Verification and Identification: Learning User Style,” presents a two-stream CNN for learning gesture style from a set of training gestures. Chapter 8 improvises the gender classification problem by deploying CNN architecture. Chapter 9, “Gender Classification from NIR Iris Images Using Deep Learning,” first attempts an unsupervised training stage with restricted Boltzmann machines (RBMs), followed by a supervised training stage with CNN leading to improved accuracy figures. Chapter 10, “Deep Learning for Tattoo Recognition,” shows how tattoo analysis can help identify ideological affiliations or gang traits.

Part 4, “Deep Learning for Biometrics Security and Protection,” contains two chapters. Chapter 11, “Learning Representations for Cryptographic Hash Based Face Template Protection” presents a CNN for learning representations that could benefit template security and extending matching accuracy. Chapter 12, “Deep Triplet Embedding Representations for Liveness Detection,” attempts to use a deep learning approach for distinguishing live or genuine fingerprints from fake ones.

This book, which covers different deep learning neural architectures for solving an extended set of problems in the area of biometrics, is sure to catch the attention of scholars and researchers working in the field.