Feature selection (FS) tackles the challenge of identifying system descriptors that best predict a given outcome. Unlike other dimensionality reduction methods, FS preserves the original meaning of features. This approach has proven successful in various C3AI applications, dealing with datasets containing an extensive number of features, which might be impractical to model and process (e.g., large-scale image analysis, text processing and web content classification), and which must retain feature semantics.

This presentation will delve into the development and application of approximate FS mechanisms based on rough and fuzzy-rough theories. These techniques offer an effective means to reduce imprecisely described data without relying on user-supplied information. Specifically, fuzzy-rough feature selection (FRFS) accommodates discrete and real-valued noisy data, making it suitable for both regression and classification. The only additional information required is the fuzzy partition for each feature, which can be automatically derived from available domain data. FRFS has been shown to be a powerful technique for semantics-preserving data dimensionality reduction. 

Beginning with an overview of the general background of FS, this talk will initially cover the rough-set-based approach before delving into FRFS and its real-world applications. The presentation will conclude with an outline of opportunities for further development.

The adversarial impact of the Covid-19 pandemic has created a health crisis globally all over the world. This unprecedented crisis forced people to lockdown and changed almost every aspect of the regular activities of the people. Thus, the pandemic is also impacting everyone physically, mentally, and economically, and it, therefore, is paramount to analyze and understand emotional responses during the crisis affecting mental health. Negative emotional responses at fine-grained labels like anger and fear during the crisis might also lead to irreversible socio-economic damages. In this talk, I will discuss machine learning models trained using manually labeled data to detect various emotions at fine-grained labels in the Covid-19 tweets automatically. A custom Q&A RoBERTa model to extract phrases from the tweets that are primarily responsible for the corresponding emotions will also be discussed. I will also present historical emotion analysis using COVID-19 tweets. Further, I will further discuss deep learning models leveraging the pre-trained BERT-base to detect the political ideology from the tweets for political polarization analysis. In addition, I will present some analysis on Hate and Offensive tweets by learning at fine-grain levels.