Title: Novel large-scale machine learning algorithm
Speaker: Dr. Mudasir Ahmad Ganaie, Postdoctoral Research Fellow, Department of Robotics, University of Michigan, Ann Arbor, USA.
Day, Date and Time: Monday, July 17, 2023, 10 AM to 11 AM (Indian time), Online
Abstract: With the explosive growth in technology, the amount and the variety of data has grown tremendously leading to new challenges in classification scenarios. Parallel hyperplane classifiers such as support vector machine (SVM) which was considered one of the most popular classification paradigm in Machine Learning owing to its strong mathematical background, has lately faced criticism due to its limitations such as unscalability, high time complexity and sensitivity to feature and label noise and poor performance in class imbalance problems. Over the past decade, several advancements have been made in the form of non-parallel hyperplane models such as twin SVM and non-parallel SVM which led to significant improvements in terms of fast learning speed, ease of implementation and ability to capture diversity among classes. These models have attracted considerable research attention due to promising results shown in the various real-world applications including Image Retrieval, Computer Vision, Financial Regression, Biomedical Analysis etc. However, there have emerged new challenges along with the existing ones such as high dimensionality in kernel implementations, need for large training data and sensitivity to outliers and bias towards the majority class samples in class imbalance problems. There is, thus, a need to improve upon these methods and devise new ones to tackle the aforementioned limitations. In the talk, I will present a novel machine learning algorithm formulated to handle large scale problems and class imbalance issues.
Speaker Bio: Mudasir Ahmad Ganaie is currently working as a Postdoctoral Research Fellow at the Department of Robotics, University of Michigan, Ann Arbor, USA. He received the Ph.D degree from Indian Institute of Technology Indore, India. Prior to that, he received the M.Tech in Computer Science and Engineering from Aligarh Muslim University, Aligarh, India. His research interests include support vector machines, machine learning, deep learning, applications to Alzheimer's disease and dementia. He has published over 26 refereed journal papers of international repute. His publications have over 951 citations with h-index 15 (Google Scholar, July 2023). He is the recipient of the 2023 IIT Indore Best Research Paper Award, and 29th ICONIP 2022 Best Research Paper Award.