Tutorial/Keynotes

Body: 

  SIRS'23 Speaker:  Dr. S. R. Mahadeva Prasanna, Professor, Dept. of Electrical Engineering, Indian Institute of Technology Dharwad, Dharwad, India

Title of the Talk: Nonlinear Speech Processing by Deep Learning Biography: Dr. S. R. Mahadeva Prasanna is a Professor of Electronics and Communication Engineering at the Indian Institute of Technology, Dharwad. With a lineage deeply rooted in education, his profound enthusiasm for teaching is reflected in his pedagogy and remarkable student feedback. His pivotal contributions extend to the developmental strides of both IIT Guwahati and the establishment of IIT Dharwad. Driven by innovation, he introduced essential courses like speech processing, neural networks, deep learning, and more, tailored effectively to cater to students from diverse academic backgrounds. This pioneering approach has facilitated students in securing admissions to prestigious institutions nationally and internationally. Beyond academia, Dr. Prasanna's commitment to research and innovation is commendable. He spearheads research groups and actively fosters collaborations between academia and industry. Notably, he mentors faculty from various engineering colleges in teaching methodologies, research, and project proposal composition. His initiatives encompass a spectrum of activities, including faculty development programs, workshops, conferences, invited talks, and cultural events, enriching both the campus community and external institutions.

Abstract: Speech processing is the forefront application area of signal processing. Most developments in digital signal processing are directly applied to speech processing. Thus signal processing and speech processing went hand-in-hand for the last many decades. Before the advent of deep learning, the feature extraction by signal processing played a major role in deciding the performance of pattern recognition systems. For more than a decade now, deep learning provided alternative ways of development of speech technologies with much superior performance. The contributions may be viewed under two vertices, representation learning doing nonlinear signal processing for feature extraction and machine learning for modeling pattern information using much more data. This talk will present some of the interesting results of deep learning as nonlinear signal processing that has delivered speech technologies with human level performance.

SIRS'23 Speaker: Dr. Sri Krishnan, Professor of Electrical, Computer Engineering, and Biomedical Engineering, Toronto Metropolitan University, Toronto, Canada

Title of the Talk: Biomedical Signal Analysis and Digital Health Biography: Sri Krishnan joined Toronto Metropolitan University, Toronto, Canada in 1999, and he is now a Professor of Electrical, Computer Engineering, and Biomedical Engineering. Sri Krishnan's research interests are in biomedical signal analysis, audio signal analysis, and explainable machine learning. He is a Fellow of the Canadian Academy of Engineering. From 2007 to 2017 he was a Canada Research Chair in Biomedical Signal Analysis.  Sri Krishnan is a recipient of the Outstanding Canadian Biomedical Engineer Award, Achievement in Innovation Award from Innovate Calgary, Sarwan Sahota Distinguished Scholar Award,  Young Engineer Achievement Award from Engineers Canada, New Pioneers Award in Science and Technology, and Exemplary Service Award from the IEEE Toronto Section.

Abstract: Biomedical data possess dynamic and complex characteristics that need to be processed using adaptive and advanced algorithms, and digital tools for data-driven decision support systems (DSS) and computer-aided diagnosis (CAD). The role of machine learning and artificial intelligence (AI) holds great promise and significance in designing proactive digital healthcare DSS and CAD. To ensure trustworthy and fair results, AI techniques need explainability to both domain experts and end users. In this talk, the process of explainable AI will be elaborated with some case study research projects being done at the Signal Analysis Research Lab at Toronto Metropolitan University, Canada.

Speaker: Karthik Sivarama Krishnan, Senior AI Software Engineer (Research), Gen Nine Inc, USA

Title of the Tutorial: Countering Audio Deepfakes: Introduction to Multi-Feature Audio Authenticity Network (MFAAN) (90 minutes)
 
Biography:  Karthik Sivarama Krishnan is a devoted researcher with an extensive background in signal processing and deep learning. His recent work on the Multi-Feature Audio Authenticity Network (MFAAN) has garnered attention for its novel approach to tackling the pervasive issue of audio Deepfakes. Karthik's commitment to advancing technology for societal benefit is evident in his consistent contribution to both academic research and practical, real-world solutions.
 
Abstract: This tutorial will delve into the burgeoning field of audio deepfakes, elucidating the challenges and potential consequences of unchecked audio manipulations. It will introduce attendees to the Multi-Feature Audio Authenticity Network (MFAAN), a cutting-edge model adept at discerning genuine recordings from manipulated counterparts. Through an exploration of MFAAN’s intricate architecture and multifaceted design, participants will gain insight into leveraging diverse audio features for enhanced deepfake detection.
 
Coverage: 
  1. Overview of Audio Deepfakes: Understanding the concept and the technology behind audio Deepfakes.
  2. Challenges in Detecting Audio Deepfakes: Discussion on why traditional techniques fall short.
  3. Introduction to MFAAN: Exploring the architecture, principles, and motivations behind MFAAN.
  4. Working of MFAAN: Delve into how MFAAN utilizes MFCC, LFCC, and Chroma-STFT for effective audio Deepfake detection.
  5. Hands-on Demonstration: Walkthrough of implementing MFAAN, showcasing its efficacy in real-world scenarios.
  6. Q&A and Discussion: Address queries and foster a dialogue on future advancements and potential refinements in audio Deepfake detection.
Prerequisites:
  • Basic understanding of deep learning concepts.
  • Familiarity with audio processing and feature extraction would be advantageous.
  • Prior exposure to the PyTorch framework is helpful but not mandatory.
Speaker: Dr. Jaya Dofe, Assistant Professor, Dept. of Electrical and Computer Engineering, California State University, Fullerton, California, USA

Title of the Tutorial: Securing the Vertical Frontier: Advancing Hardware Security through 3D Integration (60 minutes)
 
Biography:  Dr. Dofe holds a Master of Science and a Ph.D. in Electrical and Computer Engineering, conferred by the University of New Hampshire in 2015 and 2018, respectively. She is an assistant professor in the Department of Electrical and Computer Engineering at California State University, Fullerton, California, USA. Her research is primarily centered around hardware security, encompassing areas such as design obfuscation, the analysis of encryption algorithms through side channels, fault attack investigation, and the exploration of emerging technologies, including 3D hardware security. Beyond her technical pursuits, she has a keen interest in research related to engineering education, particularly focusing on active learning and equitable pedagogy. Dr. Dofe actively contributes to the academic community as a technical committee member for various IEEE conferences, including SOCC, ISVLSI, ISQED, iSES, VDAT, and VDEC. She has also chaired panels to support women in engineering at events like the 7th IEEE International Symposium on Smart Electronic Systems and the International IoT conference. Additionally, she plays a significant role as a steering member and general chair for the Workshop for Women in Hardware and Systems Security (WISE). Her commitment to the field extends to serving as a reviewer for several prestigious conferences and journals. She has received an Excellence in Teaching award from CSUF for her dedication to inclusive teaching.
 
Abstract: As the demand for more compact and powerful integrated circuits (ICs) continues to grow, three-dimensional (3D) Integration has emerged as a promising technological advancement. Unlike traditional two-dimensional (2D) chips, 3D Integration opens the door to exciting possibilities in computing platforms, enabling high-performance processors and unprecedented computational density. It presents opportunities to improve performance, functionality, and power efficiency to meet present and future design requirements, aligning with More Moore and More-than-Moore objectives. However, while 3D Integration holds great promise, it also has unique challenges in securing 3D ICs. This tutorial thoroughly explores the innovative potential of 3D Integration for enhancing security mechanisms. It delves into how this technology can be leveraged to strengthen hardware security, providing robust protection against evolving threats. It also discusses potential security vulnerabilities that may arise in 3D ICs, emphasizing the importance of thoughtful consideration and proactive measures to address these emerging issues effectively.
Speaker: Dr. Sanjay Adiwal, Centre for Development of Advanced Computing (C-DAC), Bengaluru

Title of the Tutorial: Demystifying DNS Abuses - Tunneling and DDoS Amplification techniques and mitigations(60 minutes)
 
Biography:  Dr. Sanjay Adiwal has over 18 years of professional experience and has a keen interest for research in the area Cyber Security for IT and IoT.  He holds a Bachelor's degree in Computer Science, a Master's degree in Computer Application. Obtained PhD in Cyber Security from NITK Surathkal Mangalore. Currently he is working in CDAC Bangalore as Joint Director. Under the current role as VAPT Project Incharge, he is responsible for overseeing the end-to-end execution of VAPT assessments, ensuring that client's digital assets are robustly protected against emerging cyber threats. His leadership and technical acumen have led to successful outcomes in numerous high-stakes VAPT engagements. He is also leading the team which manages Data Center of CDAC Bangalore. Prior to CDAC he was working for Indian Institute of Science (IISc) Bangalore as system administrator trainee. Dr. Sanjay is also highly regarded cybersecurity speaker and trainer, known for his deep expertise and dynamic presentations in the field of cybersecurity. He has conducted knowledge dissemination programs/workshops across various cities in India under the project - PKI Public Key Infrastructure Body of Knowledge. And conducted more than 100 workshops and seminars across various cities in the country under Information Security Education and Awareness (ISEA) project of India Govt taken by ICT under the initiative of CDAC. teaching.
 
Abstract:  The Domain Name System (DNS) serves as the backbone of the internet, translating user-friendly domain names into machine-readable IP addresses. However, this critical infrastructure is not immune to abuse. This tutorial, "DNS Abuses - Tunneling and DDoS Amplification Techniques and Mitigations," delves into the darker aspects of DNS, exploring two prevalent threats: DNS tunneling and DDoS amplification attacks. DNS tunneling involves the covert encapsulation of non-DNS traffic within DNS packets, enabling data exfiltration and command and control channels for cybercriminals. DDoS amplification attacks leverage vulnerabilities in DNS to amplify malicious traffic, wreaking havoc on targeted networks. In this tutorial, I will provide a comprehensive understanding of these threats, dissecting their mechanics and showcasing real-world examples to underline their severity. By the end of this tutorial, participants will possess the knowledge and skills needed to safeguard their networks and systems against DNS abuses, preserving the integrity and availability of their online infrastructure.