Lecture on "Dimensionality Reduction"
Mrs. L. V. Rajani Kumari (Assistant Professor, VNR VJIET) was the resource person for the day to deliver a lecture on “Dimensionality Reduction Techniques” to help students realize the problems with High Dimensional Data, Presence of noise, and the need of reducing dimensions using certain techniques, along with examples and use-cases for better understanding.
- At the very beginning of the presentation, there was a recap on general classification from previous sessions where topics like Feature Extraction, Ensemble Learning, cross-validation, etc. in order to acquire the best performance of a Machine Learning model.
- Then, the issues like the presence of noise in data, correlation among features, curse of dimensionality, etc. were discussed, how general classification techniques become ineffective, and how these issues degrade the model performance.
- A common misconception on the difference between the terms ‘More data’ and ‘More features’ was cleared, for further progress of the session.
- Advantages of Dimensionality Reduction Techniques were discussed and how it is useful in obtaining optimal number of features, explained along with examples.
- Two popular techniques for Dimensionality Reduction were discussed. They are:
- PCA (Principal Component Analysis)
- LDA (Linear Discriminant Analysis)
- Each of the techniques was explained with step-by-step mathematical computations, where topics like Linear Algebra, Matrices, Eigenvalues and Eigenvectors, Covariance of a Matrix, etc. were covered, along with numerical examples.
- A use-case of Principal Component Analysis was covered in a real-time application, i.e. in Electrical Activity of Heart by analyzing ECG Signal data. The Program of the same, was demonstrated on MATLAB.
Date and Time
- Date: 03 Jul 2020
- Time: 02:00 PM to 04:00 PM
- All times are (GMT+05:30) Asia/Calcutta
- Add Event to Calendar
L.V Rajani Kumari of Assistant Professor, Department of Electronics & Communication Engineering, VNR VJIET
Lecture on Dimensionality Reduction
Mrs. L. V. Rajani Kumari has a teaching experience of over 5 years. She has pursued an M.Tech in Embedded Systems. She is currently pursuing her Ph.D in Biomedical Signal Processing. Also, she published 2 journal papers on use-cases of Machine Learning and Artificial Neural Networks in Signal Processing.
In this session, the resource person presented a brief introduction, on what is dimensionality reduction, problems with high dimensional data, and the need of Dimensionality Reduction Techniques in the Feature Extraction process. Some of the well-known techniques were keenly discussed and demonstrated on MATLAB.
About 28 people attended the webinar and gained knowledge on Dimensionality Reduction Techniques for Data Compression, Noise Filtering etc. The speaker executed each and every program in detail to benefit the students and clarify their doubts regarding the mathematical approaches taught in PCA and LDA. This session benefitted both beginners and professionals which helps them in better understanding of algorithms used for Dimensionality Reduction.