Machine Learning and Symbolic AI - the Best of Both Worlds

June 14, 2021 by Victor de Beur

Dr. Mehwish Alam is a Post-Doctoral Researcher/Senior Researcher at FIZ Karlsruhe. She is one of this year’s Research and Innovation Chairs. The focus of her research is to apply or develop Data Mining, Machine Learning/Deep Learning techniques for Semantic Web and Text Processing. 

SEMANTiCS: One of your areas of expertise is explainability in symbolic artificial intelligence systems. What are the challenges in this special branch of AI research?

Mehwish Alam: Explainability is one of the important fields in Artificial Intelligence that targets the need for trustworthiness on the predictions made by these AI-based systems. This trustworthiness should be ensured at each step, i.e., starting from the step of data preparation to the predictions made by the system. Symbolic Representations have been used for explaining the predictions or the recommendations of the algorithms in a post-hoc manner. These systems also utilize Symbolic Representations as part of their framework/architecture. However, with the advent of sub-symbolic methods self-explanation has become more critical.

Want to learn more about Explainability? Register for SEMANTiCS 2021!

SEMANTiCS: Especially the marriage of AI and Knowledge Graphs stimulated research and innovation in the sector, bringing various strands of AI research together. What will be the role of KGs in AI systems of the future - hype or hope?

Mehwish Alam: Subsymbolic AI is data-hungry, however, methods developed so far for Knowledge Representation and Reasoning have difficulties in processing large amounts of data. In the era of big data, machine learning and deep learning play a vital role in enabling data analysis. However, combining Machine Learning and Symbolic AI might help in bringing together the best of both worlds and developing more efficient systems.

SEMANTiCS: Together with Paul Groth from The University of Amsterdam, you are this year’s Research and Innovation Chair. Thinking ahead of the obvious next developments, is there a topic or development which you want to see discussed?

Mehwish Alam: I think by now you already know what would interest me a lot. Since Semantics is the venue where researchers and industries come together, it would be very amazing to see the applicability of Explainable AI as well as Symbolic AI and Machine Learning in the industry and real-world applications such as materials science, sensor data, etc.

About Mehwish Alam:

Dr. Mehwish Alam is a Post-Doctoral Researcher/Senior Researcher at FIZ Karlsruhe – Leibniz Institute for Information Infrastructure and Karlsruhe Institute of Technology (KIT), Institute of Applied Informatics and Formal Description Methods (AIFB) with Prof. Dr. Harald Sack. Before that she has conducted her Post-Doctoral Research at Laboratoire d’Informatique de Paris-Nord (LIPN), Paris, France (2016-2017) and Consiglio Nazionale delle Ricerche (CNR), Rome, Italy (2017-2019) with Prof. Dr. Aldo Gangemi . The focus of her research is to apply or develop Data Mining, Machine Learning/Deep Learning techniques for Semantic Web and Text Processing. Before her Post-Doc, she was PhD student in the field of Computer Science (2011-2015) in LORIA, INRIA, Nancy Grand-Est, France. Her PhD thesis was supervised by Dr. Amedeo Napoli and Dr. Malika-Smail Tabbone. She completed her prestigious Erasmus Mundus Research Masters in Language Communication and Technology in the discipline of Natural Language Processing. She completed her Masters in collaboration with University of Groningen, Groningen, the Netherlands and University of Nancy 2, Nancy, France. Her publications are accessible on DBLP and Google Scholar.

She has served as a proceedings chair as well as publicity chair for the ESWC'18, Poster and Demo Session chair in SEMANTiCS Conference 2019, Program Chair of International Conference in Conceptual Structures (ICCS'19), Assistant Tutor in International Semantic Web Summer School (ISWS'18). She has also been one of the organizers of the Workshop on Deep Learning for Knowledge Graphs (DL4KG'19) at ESWC 2019 and Fourth International Workshop at ESWC 2018 on Sentic Computing, Sentiment Analysis, Opinion mining and Emotion Detection. She is part of the program committee of major conferences such as IJCAI, AAAI, EKAW, ESWC etc. and served as a reviewer in many journals and workshops.