Deep learning and its adversarial robustness: A brief introduction

TY - CHAP

T1 - Deep learning and its adversarial robustness

T2 - A brief introduction

AU - Wang, Fu

AU - Zhang, Chi

AU - Xu, Peipei

AU - Ruan, Wenjie

N1 - © 2022 World Scientific Publishing Company

PY - 2022/9/1

Y1 - 2022/9/1

N2 - Deep learning, one of the most remarkable techniques in computational intelligence, has become increasingly popular and powerful in recent years. In this chapter, we, first of all revisit the history of deep learning and then introduce two typical deep learning models including convolutional neural networks (CNNs) and recurrent neural networks (RNNs).After that, we present how the deep learning models are trained and introduce currently popular deep learning libraries and frameworks. Then we focus primarily on a newly emerged research direction in deep learning—adversarial robustness. Finally, we show some applications and point out some challenges of deep learning. This chapter cannot exhaustively cover every aspect of deep learning. Instead, it gives a short introduction to deep learning and its adversarial robustness, and provides a taste of what deep learning is how to train a neural network, and why deep learning is vulnerable to adversarial attacks, and how to evaluate its robustness.

AB - Deep learning, one of the most remarkable techniques in computational intelligence, has become increasingly popular and powerful in recent years. In this chapter, we, first of all revisit the history of deep learning and then introduce two typical deep learning models including convolutional neural networks (CNNs) and recurrent neural networks (RNNs).After that, we present how the deep learning models are trained and introduce currently popular deep learning libraries and frameworks. Then we focus primarily on a newly emerged research direction in deep learning—adversarial robustness. Finally, we show some applications and point out some challenges of deep learning. This chapter cannot exhaustively cover every aspect of deep learning. Instead, it gives a short introduction to deep learning and its adversarial robustness, and provides a taste of what deep learning is how to train a neural network, and why deep learning is vulnerable to adversarial attacks, and how to evaluate its robustness.

UR - https://www.worldscientific.com/doi/abs/10.1142/9789811247323_0013

U2 - 10.1142/9789811247323_0013

DO - 10.1142/9789811247323_0013

M3 - Chapter

SN - 9789811247323

SN - 9789811245145

VL - 2

SP - 547

EP - 584

BT - Handbook on Computer Learning and Intelligence

A2 - Angelov, Plamen Parvanov

PB - World Scientific

ER -