Speech Technology has advanced significantly beyond general speech recognition for voice command and telephony applications. Today, the emergence of BIG DATA, Machine Learning, as well as voice enabled speech systems have required the need for effective voice capture and automatic speech/speaker recognition. The ability to employ speech and language technology to assess human-to-human interactions is opening up new research paradigms which can have a profound impact on assessing human interaction. In this talk, we will focus on BIG DATA audio processing relating to the APOLLO lunar missions. ML based technology advancements include automatic audio diarization and speaker recognition for audio streams which include multi-tracks, speakers, and environments. CRSS-UTDallas built a recovery solution for lost 30-track audio tapes from NASA Apollo-11, resulting in a massive multi-track audio processing (19,000hrs) of data. Recent additional support from NSF will allow for the recovery and organization of an additional 150,000hrs of mission data to be shared with the communities of: (i) speech/language technology, (ii) STEM/science and team-based researchers, and (iii) education/historical/archiving specialists.
John H.L. Hansen, received Ph.D. & M.S. degrees from Georgia Institute of Technology, and B.S.E.E. degree from Rutgers Univ. He joined Univ. of Texas at Dallas (UTDallas) in 2005, where he is Associate Dean for Research, Professor of Electrical & Computer Engineering, Distinguished Univ. Chair in Telecommunications Engineering, and holds a joint appointment in School of Behavioral & Brain Sciences (Speech & Hearing). At UTDallas, he established Center for Robust Speech Systems (CRSS). He is an ISCA Fellow, IEEE Fellow, past
Member and TC-Chair of IEEE Signal Proc. Society, Speech & Language Proc. Tech. Comm.(SLTC), and Technical Advisor to U.S. Delegate for NATO (IST/TG-01). He currently serves as ISCA President. He has supervised 92 PhD/MS thesis candidates, was recipient of 2020 UT-Dallas Provost’s Award for Grad. Research Mentoring, 2005 Univ. Colorado Teacher Recognition Award, and author/co-author of +750 journal/conference papers in the field of speech/language/hearing processing & technology. His has over 20,000 citations. He served as General Chair for Interspeech-2002, Co-Organizer and Tech. Chair for IEEE ICASSP-2010, and Co-General Chair and Organizer for IEEE Workshop on Spoken Language Technology (SLT-2014) (Lake Tahoe, NV). He is serving as Co-Chair for ISCA INTERSPEECH-2022, and Tech. Chair for IEEE ICASSP-2024.
Title:A Machine Learning Based Framework for Validating Code Clones from Big Code
Chanchal K. Roy, Ph. D
Co-Director of Software Research Lab
Professor, Department of Software Engineering /Computer Science
A code clone is a pair of code fragments, within or between software systems that are similar. Since code clones often negatively impact the maintainability of a software system, several code clone detection techniques and tools including those Big data clone detectors have been proposed. However, the clone detection tools are not always perfect and their clone detection reports often contain an unknown number of false positives or irrelevant clones from specific project management or user perspective. Such issues even become more crucial for Big Data clone detectors where a clone detection tool detects millions of clones from Big Data software repositories. A clone detector may detect thousands of false positive clones from Big Code which is impossible to validate by human, let alone doing it accurately. Furthermore, to detect all possible similar source code patterns in general, the clone detection tools work on the syntax level while lacking user-specific preferences. This often means the clones must be manually inspected before analysis in order to remove those false positives from consideration. This manual clone validation effort is very time-consuming and often error-prone even for an accurate clone detection tool. In this talk, we propose a machine learning approach for automating the validation process. First, a training dataset is built by taking code clones from several clone detection tools for different subject systems and then manually validating those clones. Second, the trained model is further refined with millions of validated clone pairs from BigCloneBench Dataset, a clone benchmark with over 8.5 million clone pairs from 25K Java projects. Third, several features are extracted from those clones to train the machine learning model by the proposed approach. The trained algorithm is then used to automatically validate clones without human inspection. Thus the proposed approach can be used to remove the false positive clones from the detection results, automatically evaluate the precision of any clone detectors for any given set of datasets, evaluate existing clone benchmark datasets, or even be used to build new clone benchmarks and datasets with minimum effort. In an experiment with clones detected by several clone detectors in several different software systems, we found our approach has an accuracy of up to 87.4% when compared against the manual validation by multiple expert judges. The proposed method also shows better results in several comparative studies with the existing related approaches for clone classification. I will also talk about how classical clone detection tools could be scaled for Big Data with the validation framework. I will also talk about a couple of other applications of machine learning and Big Data Analytics in Software Engineering, in particular I will talk about the application of Big Data analytics for bug localization and concept location.
Chanchal K. Roy is Co-Director of Software Research Lab and Professor of Software Engineering/Computer Science at the University of Saskatchewan, Canada. He is the lead and Program Director of an NSERC CREATE graduate program on Software Analytics Research and a co-lead of the Data Management and Repository group of an NSERC Canada First Research Excellence Fund (CFREF) on Food security. As the co-author of the widely used NICAD code clone detection system, he has published more than 170 refereed publications, with many of them in premier software engineering conferences and journals that have been cited more than 6,000 times. Dr. Roy works in the broad area of software engineering, with particular emphasis on software clone detection and management, software evolution and maintenance, recommender systems in software engineering, automated software debugging, and big data analytics in software engineering. His contributions to the software maintenance community, and particularly to the software clones community, have been highly influential, winning Most Influential Paper awards at both SANER 2018 and ICPC 2018. He has been recognized with the New Scientist Research Award of the College of Arts and Science of the University of Saskatchewan and the University wide New Researcher Award. He is one of three Canadian computer scientists honoured with a prestigious award for young researchers, a 2018 Outstanding Young Computer Science Researcher Award by CS-Can/Info-Can, a national, non-profit society dedicated to representing all aspects of computer science and the interests of the discipline across Canada. Dr. Roy was a vision keynote speaker at WCRE/CSMR 2014 on software clones, and a keynote speaker at both IWSC 2018 and IEEE R10HTC 2018. He serves widely on the program committees of major software engineering conferences such as ICSE, ASE, ICSME, SANER, MSR, ICPC and SCAM, and has been regular reviewers of the major journals in Software Engineering. He served (has been serving) as chairs and/or program committee members in most of the conferences of his area including General Chair for ICPC 2014, SCAM 2019, IWSC 2015 and Program Co-chairs for ICPC 2018 and IWSC 2012. He has attracted over $4M in external funding since joining the USask, including an NSERC Discovery Accelerator Supplement Grant, NSERC CREATE grant and leading major roles in two CFREF grants in Food Security and Water Security. Dr. Roy’s recent work on a new way of searching Stack Overflow was featured in Stack Overflow blogs which then subsequently was featured in most of the major tech news websites and blogs such as ACM Tech news, TechRepublic, Hacker News, SD Times, and reddit.
Title: Neurobiology and Deep Learning: First Principles and Future Applications
Paul Watters, Ph. D
Adjunct Professor, Comp Science and Information Technology
Computational models of learning and neurobiology have a long and complex history, starting with the McCulloch-Pitts neuron, through to perceptrons, backpropagation, and most recently, deep learning. In all cases, computational models have interpreted the structure and function in a way which has not only modelled cellular responses to external stimuli, but has algorithmically achieved extraordinary success in implementing real learning functions. In this talk, I will give some insights and future possible areas of investigation to further drive algorithmic development in learning, but also to identify cognate areas in cognitive neuroscience that could be further inspired by deep learning.Biography:
Professor Watters managed his first cybersecurity incident in 1993 while working as a UNIX system administrator at the University of Newcastle. He worked in technical roles for many years, including as a system administrator and software engineer for IT companies with clients such as Oracle Australia, Fuji Xerox, the National Office of the Information Economy (NOIE), Qantas, Kip McGrath Education Centres, the NSW Police Service, Pickles Auctions, Sydney Fish Markets and the Universities Admissions Centre (UAC). He also worked as a researcher at the Mental Health Research Institute of Victoria, CSIRO, Macquarie University, the Medical Research Council (UK), the University of Ballarat, Massey University and La Trobe University. His R&D client list includes Westpac, National Australia Bank, Australian Federal Police, Attorney General’s Department, the Motion Picture Association, and End Child Prostitution and Trafficking (ECPAT).
Professor Watters has published a number of top-selling IT books on cyber security, intelligence, and system administration with McGraw Hill, O’Reilly & Associates and Springer. Professor Watters is one of the world’s leading cyber security researchers, with his work being cited more than 4,310 times by other researchers, with a h-index of 29, and an i10-index of 81. A full listing of his research papers spanning malware analysis, phishing, forensics, Child Sex Abuse Material (CSAM) prevention, cyber intelligence, anti-piracy and others can be found on Google Scholar.
Title: Activity Analysis: Healthcare and Related Perspectives
Md Atiqur Rahman Ahad, Ph. D
Department of Electrical & Electronic Engineering
University of Dhaka, Bangladesh
Specially appointed Associate Professor at Osaka University, Japan
Human Activity Recognition (HAR) are explored in video-based computer vision domain and sensor-based ubiquitous research areas. Vision-based human action or activity recognition approaches are based on RGB video sequences, or depth maps, or from skeleton data – taken from normal video cameras or depth cameras. On the other hand, sensor-based activity recognition methods are basically based on the data collected from wearable sensors having accelerometer, gyroscope, or so on. There are numerous applications on HAR, however, the healthcare, elderly support, and related applications become very important arenas with huge social and financial impact. Due to the advent of various IoT sensors, it becomes more competitive as well as easier to explore different applications. The keynote will cover HAR approaches in both video-based and sensor-based domains, highlight healthcare perspectives and methods. The presentation will be based on the books as follows:
1. Md Atiqur Rahman Ahad, Anindya Das Antar, and Masud Ahmed, “IoT Sensor-Based Activity Recognition – Human Activity Recognition”, Springer Nature Switzerland AG, 2020.
2. Md Atiqur Rahman Ahad, Upal Mahbub, and Tauhidur Rahman, “Contactless Human Activity Analysis, Publisher: Springer Nature Switzerland AG, appear in 2020.
3. Md Atiqur Rahman Ahad, “Motion History Images for Action Recognition and Understanding”, Springer, 2013.
4. Md Atiqur Rahman Ahad, “Computer Vision and Action Recognition: A Guide for Image Processing and Computer Vision Community for Action Understanding”, available in Springer, 2011.Biography:
Md Atiqur Rahman Ahad, Senior Member, IEEE, is a Professor of Electrical & Electronic Engineering, University of Dhaka (DU). He is currently working as specially appointed Associate Professor at Osaka University, Japan. He works on computer vision, imaging, IoT, healthcare, etc. He did B.Sc.(honors) [1st class 1st position] & Masters [1st class 2nd position] from the Dept. of Applied Physics & Electronics, DU; Masters from the School of Computer Science & Engineering, University of New South Wales; and PhD from the Faculty of Engineering, Kyushu Institute of Technology [KIT]. He was awarded JSPS Postdoctoral Fellowship, prestigious UGC Award 2016, and a no. of awards/scholarships. He was a Visiting Researcher at KIT. He published 3 books (available in Springer), few edited books, and a few book chapters. He has published 140+ journals and conference papers. He has received 20+ international awards in various conference/journal/society. Ahad was invited as keynote/invited speakers about 60 times in different conferences/universities. He has established several international MOU/collaborations (e.g., Clemson University, University of Hyogo, RCCIIT, Fukuoka Women University, Kyushu University, etc.). Prof. Ahad has been involved with some academic & editorial activities: e.g., Editorial Board Member, Scientific Reports, Nature; Associate Editor, Frontiers in Computer Science; Editor-in-Chief: Int. J. of Computer Vision & Signal Processing http://cennser.org/IJCVSP. Ahad is a Member of OSA, ACM, IEEE Computer Society, IAPR, IEEE RAS, IEEE SMC, etc.
Title: Challenges and Opportunities of Big Data, Machine Learning and Speech Recognition
Big Data which evolves as a result of digital footprint, which is the consequence of the digitalization of various daily activities of human beings. However, this Big data is empirical in nature and hence, they are fundamentally limited by empirical science. Therefore, to make sense out of this huge amount of data, the amalgamation of analytical or rationalistic approach with BiG Data is considered as fundamental nowadays. Artificial Intelligence (AI) with its machine learning component can be seen as the appropriate method to sense Big Data, enabling us to solve real world problems. In other ways, we can say machine learning methods would embed intelligence within Big Data, facilitating to become more valuable and precise. Since this digital footprint comes in most of the cases in real time, machine learning methods would allow capturing automated insights of these data. Human speech recognition can be considered such an exemplar real world problem, where we can get a digital footprint because of the real time interaction between man and machine. By using various machine learning approaches, especially deep learning, we can automatically capture the insights of such data. An example of such insight may be considered as human emotion recognition in real time. Therefore, in this talk the results of human speech recognition will also be demonstrated where variants of deep learning approaches were explored. However, there exist issues of how such machine learning approaches can be made more transparent with the inclusion of knowledge-based approaches. Such an investigation will allow us to understand the requisition of trade-offs required between the accuracy and the explainability of the future AI.
Mohammad Shahadat Hossain is a Professor of Computer Science and Engineering at the University of Chittagong, Bangladesh since 2007. He is the first professor of Computer Science and Engineering of this region of Bangladesh. He did both his MPhil and PhD in Computation from the University of Manchester Institute of Science and Technology (UMIST), UK in 1999 and 2002 respectively. He awarded prestigious Commonwealth Academic Staff Fellowship in 2009 through a rigorous academic evaluation at The University of Manchester, UK. He also awarded prestigious Tyndall Visiting Fellowship in 2006. In 2011 and 2013, Professor Hossain awarded prestigious European Commission sponsored Erasmus Mundus Academic Staff Fellowship at the University of Aalborg, Denmark. He is also the holder of PERCCOM’s (Pervasive Computing and Mobile Communication for Sustainable Development) Scholar Scholarship as a Visiting Professor, which is sponsored by the European Commission; starting from 2014 to till-to-date He successfully completed a number of research projects. Recently, he awarded prestigious Swedish Research Council grant for the project entitled “A Belief Rule Based DSS to Assess Flood Risks using Wireless Sensor Networks”, where he is working as a Foreign Node Leader. His current research area includes the novel idea of sustainable computing which combines pervasive computing with belief rule based expert systems. Investigation of pragmatic software development tools and methods for Information Systems in general and GIS, in particular, is also his area of research. He is continuing his research in other adventurous areas of computing such as health informatics, affective computing, e-government, smart cities, Internet of Things (IoT) and philosophy of computing. He has published 130 scholarly articles in the reputed international journals and conferences. He is the author of a number of books. His jointly authored book entitled “Computing Reality”, published by Aoishima Research Institute (blue ocean press) in Tokyo, Japan, contributed significantly to enrich the knowledge of computer science and indexed in the ACM Digital Library, published by the Association for Computing Machinery, as one of the important computing literature books.
Title: Recent Applications of ML+DL in Language Processing Tasks: Bengali Language Perspectives
Moshiul Hoque, Ph. D
Department of Computer Science and Engineering,
Chittagong University of Engineering and Technology, Bangladesh
In recent years, the digital Bengali text content has been growing readily on the Internet, news portals, blogs, websites, and soon due to the effortless usage of electronic gadgets. This content is creating an enormous amount of unstructured data. Therefore, it is a challenging task to organize, search or manipulate such a massive amount of unstructured data by human experts manually. However, manual processing of voluminous data into their pre-defined classes demands huge time, enormous effort and cost of money which may inaccurate, or infeasible in most cases. Thus, automatic language processing system can handle a massive amount of Bengali text data in which documents can be sorted, manipulated and organized expeditiously and competently. Automatic processing of Bengali text is the most challenging task due to the constitution of the language itself having well-off dialects and complex morphological structure. It also articulates the huge variations of subject-verb and person-tense-aspect agreements. Unavailability of standard Bengali text corpus and scarcity of resources are antecedents that make such a text categorization task very complicated. Moreover, there are no efficient tools available that have been developed till today for Bengali language processing (BLP). Therefore, there is an insistence of developing tools for BLP so that professionals, as well as the common people, can use these tools to their needs. The application of machine learning (ML) and deep learning (DL) has growing interest among BLP experts due to its potential real-world applications in various language processing tasks such as text classification, emotion classification, sentiment analysis, authorship attribution, and suspicious text detection. This talk covers the few recent developments on language processing tasks in Bengali language including corpus development, architecture, and stat-of-the art experimental result findings.Biography:
Mohammed M. Hoque is a Distinguish Professor of the Department of Computer Science & Engineering (CSE). He received Ph. D from the Dept. of Information and Computer Sciences, Saitama University, Japan in 2012. He served as a TPC Chair (IEEE r10 HTC 2017, ECCE 2019, ICREST 2021), TPC Co-chair (ICISET 2018, IEEE TenSymp 2020), Publication chair (IEEE WIECON-ECE 2018/2019, IEEE TenSymp 2020) and TPC members in several international conferences. Dr. Hoque was the Award Coordinator (2016-17), Conference Coordinator (2017-18) of IEEE Bangladesh Section and now he is serving as Vice-chair (Technical) (2018-20). Moreover, he is serving as Vice-chair (Activity) (2018-19), Award Coordinator (2017-18) of IEEE Computer Society Bangladesh Chapter and Educational Activity Coordinator (2017-18), IEEE RAS, Bangladesh Chapter respectively. He is the founding Director of CUET NLP Lab. He published more than 115 publications in several International Journals, Book Chapters and Conferences. His research interests include human robot/computer interaction, computer vision, and natural language processing. Dr. Hoque is a senior member of IEEE, IEEE Computer Society, IEEE Robotics and Automation Society, IEEE Women in Engineering, IEEE Signal Processing Society and Fellow of Institute of Engineers, Bangladesh.