The Transition From White Box to Black Box: Challenges and Opportunities in Signal Processing Education

Modern engineering education is increasingly assuming an interdisciplinary character, where developments in one area almost invariably affect other areas. A prominent example is that of signal processing, which has undergone significant changes with the emergence of machine learning (ML) and deep learning (DL) in recent years. While the impact of ML/DL is clearly visible from the viewpoint of research and development as well as industrial applications, it is not immediately clear how signal processing education should evolve in terms of pedagogy and content. Hence, the main purpose of this article is to provide some insight into this aspect. In particular, we emphasize that the introduction and popularity of ML/DL, especially at the level of teaching, has provided an opportunity to bring the focus back to some of the fundamental ideas rooted in signal processing and other related fields of study.

[1]  Wolfgang Heidrich,et al.  HDR-VDP-2: a calibrated visual metric for visibility and quality predictions in all luminance conditions , 2011, ACM Trans. Graph..

[2]  Shie Mannor,et al.  The kernel recursive least-squares algorithm , 2004, IEEE Transactions on Signal Processing.

[3]  Bülent Yener,et al.  Learning filter widths of spectral decompositions with wavelets , 2018, NeurIPS.

[4]  Frans Coenen,et al.  FCNN: Fourier Convolutional Neural Networks , 2017, ECML/PKDD.

[5]  Marcel van Gerven,et al.  Explainable Deep Learning: A Field Guide for the Uninitiated , 2020, J. Artif. Intell. Res..

[6]  S LewMichael,et al.  Deep learning for visual understanding , 2016 .

[7]  Lei Zhang,et al.  Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising , 2016, IEEE Transactions on Image Processing.

[8]  Jun Yu,et al.  Deep Learning for Multimedia: Science or Technology? , 2018, ACM Multimedia.

[9]  Avanti Shrikumar,et al.  Fourier-transform-based attribution priors improve the interpretability and stability of deep learning models for genomics , 2020, bioRxiv.

[10]  Anthony G. Constantinides,et al.  Bringing Wearable Sensors into the Classroom: A Participatory Approach [SP Education] , 2018, IEEE Signal Process. Mag..

[11]  Xiaogang Wang,et al.  Deep Learning for Visual Understanding [From the Guest Editors] , 2017, IEEE Signal Process. Mag..

[12]  Harald Brendel,et al.  Creating cinematic wide gamut HDR-video for the evaluation of tone mapping operators and HDR-displays , 2014, Electronic Imaging.

[13]  Xin Li,et al.  Adversarial Examples Detection in Deep Networks with Convolutional Filter Statistics , 2016, 2017 IEEE International Conference on Computer Vision (ICCV).

[14]  Rama Chellappa.  The changing fortunes of pattern recognition and computer vision , 2016, Image Vis. Comput..

[15]  Xiaowen Dong,et al.  Graph Signal Processing for Machine Learning: A Review and New Perspectives , 2020, IEEE Signal Processing Magazine.

[16]  Junjie Wu,et al.  Multilevel Wavelet Decomposition Network for Interpretable Time Series Analysis , 2018, KDD.

[17]  Rushil Anirudh,et al.  MARGIN: Uncovering Deep Neural Networks using Graph Signal Analysis , 2017 .

[18]  Alessandro Foi,et al.  Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering , 2007, IEEE Transactions on Image Processing.

[19]  Eric C. Larson,et al.  Most apparent distortion: full-reference image quality assessment and the role of strategy , 2010, J. Electronic Imaging.

[20]  Oisín Boydell,et al.  An Overview of Deep Learning in Industry , 2020 .

[21]  Giulia Boato,et al.  Teaching Digital Signal Processing With a Challenge on Image Forensics [SP Education] , 2019, IEEE Signal Processing Magazine.

[22]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .

[23]  Rob Fergus,et al.  Visualizing and Understanding Convolutional Networks , 2013, ECCV.

[24]  Michael Rabadi,et al.  Kernel Methods for Machine Learning , 2015 .

[25]  A. M. Turing,et al.  Computing Machinery and Intelligence , 1950, The Philosophy of Artificial Intelligence.

[26]  Eric P. Xing,et al.  High-Frequency Component Helps Explain the Generalization of Convolutional Neural Networks , 2019, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[27]  Anthony G. Constantinides,et al.  On the Intrinsic Relationship Between the Least Mean Square and Kalman Filters [Lecture Notes] , 2015, IEEE Signal Processing Magazine.

[28]  J. Moura,et al.  Graph Signal Processing and Deep Learning: Convolution, Pooling, and Topology , 2020, IEEE Signal Processing Magazine.

[29]  Jonathan T. Barron,et al.  Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains , 2020, NeurIPS.