Introducing Machine Learning in Undergraduate DSP Classes

Machine Learning (ML) and Artificial Intelligence (AI) algorithms are enabling several modern smart products and devices. Furthermore, several initiatives such as smart cities and autonomous vehicles utilize AI and ML computational engines. The current and emerging applications and the growing industrial interest in AI necessitate introducing ML algorithms at the undergraduate level. In this paper, we describe a series of activities to introduce ML in undergraduate digital signal processing (DSP) classes. These activities include a computational comparative study of ML algorithms for spoken digit recognition using spectral representations of speech. We choose spectral representations and features for speech as those concepts associate with the core topics in DSP such as FFT and autoregressive spectra. Our primary objective is to introduce undergraduate DSP students to feature extraction and classification using appropriate signal analysis and ML tools. An online module on ML along with a computer exercise are developed and assigned as a semester project in the DSP class. The exercise is developed in Python and also on the online JDSP HTML5 environments. An assessment study of the modules and computer exercises are also part of this effort.

1. INTRODUCTION

Machine learning and artificial intelligence applications have grown rapidly across several disciplines, industries and cultures. Although some computer science and engineering curricula [1]-[3] include AI training, most Electrical Engineering undergraduate plans-of-study do not cover ML. On the other hand, many of the text books and survey papers have targeted graduate student audiences [4]-[6]. Several authors have argued that AI education must start early [1],[7]. In this paper, we describe and assess an effort to introduce ML in our undergraduate Digital Signal Processing (DSP) classes. This activity includes developing online teaching modules and appropriate software to support computer exercises and projects. We started this activity initially for use in our SenSIP REU site program [8] and we developed a small J-DSP module and a k-means based exercise for our class and REU summer students [9]. At Arizona State, we teach core DSP at the senior level, though we have a DSP course for non-majors [10] as well. The DSP class is offered both as face-to-face and in a flipped course format [11]-[12]. We also routinely use some DSP materials for outreach and teacher training [13] using approaches such as those described in [14],[20]. Introducing machine learning for all these activities is accomplished using a scalable approach and appropriate hands-on computer exercises. The module for the early DSP class has qualitative content and is accompanied by user-friendly J-DSP object oriented (block diagram based) software. The module for the senior level DSP class include Python and MATLAB programming [15] and therefore has a skill building component.