Karthika Vijayan

Short CV: Dr. Karthika Vijayan is a Solution Consultant at Sahaj AI Software. She has been conducting research in the field of conversational AI with voice and text data for almost a decade. Her research has been published in several journals and presented at various international conferences. Prior to joining Sahaj, Dr. Vijayan worked as a research fellow in the Department of Electrical and Computer Engineering at the National University of Singapore. She also worked as research associate in the Department of Electrical Engineering at the Indian Institute of Science, Bangalore. She obtained her PhD in Speech Signal Processing from the Department of Electrical Engineering at the Indian Institute of Technology Hyderabad in 2016. She is an active member of IEEE, ACM and APSIPA. https://www.linkedin.com/in/karthika-vijayan/

Title of Talk 1: Multilingual Conversational AI for Human Computer Interaction

Synopsis: Natural language processing (NLP) has come a long way from n-gram representations through recursive models to transformer models. Along the way, we have achieved imparting intelligence into text processing systems. In this talk, we will briefly discuss history of NLP, the development of transformer-based models for language representation, and multilingual NLP using these representational models. Utilisation of massively multilingual language representation models has made significant progress in realisation of various NLP tasks in multilingual scenario. Text processing for numerous applications like topic categorisation, sentiment analysis and information extraction for rich multilingual societies largely benefit from availability of the single representational model that can be used for NLP tasks in multiple languages.

Title of Talk 2: Disambiguated Knowledge Graphs

Synopsis: Knowledge graph provides meaningful and structured representation for information residing in unstructured data like text. It allows easier querying for information retrieval, that can be beneficial for several applications like search, question answering, information extraction and compact representation of related text data from multiple sources. In this talk, creation of knowledge graphs from complex text articles and visually rich documents will be discussed along with its prominent use-cases in natural language processing. Knowledge graph creation and completion pipelines will be explained and querying of graphs resulting in capture of information from multiple sources of documents will be showcased.

Title of Talk 3: Voice Conversational AI

Synopsis: Speech is the most natural form of human communication. So it only makes perfect sense for an AI to communicate with humans using voice. This talk will focus on the fundamentals of automatic speech recognition and speech synthesis, which form the pillars upon which voice AI is balanced upon. Automatic speech recognition (ASR) enables the AI to understand messages from its users, while, text-to-speech (TTS) synthesis facilitate the AI to talk in response to user messages. Machine learning techniques and signal processing strategies catering to speech signals are keys to build and maintain an efficient voice AI. In this talk, we will discuss the basics of speech processing, problem definition of statistical ASR and corresponding state-of-the-art solutions, and the basics of TTS synthesis and generation of naturally sounding AI voices.