Expected grade: B+
Actual grade: B+
Assessment and workload:
– 10% Programming assignment on implementing DCT (Median was 9, my score was in 25%tile, likely to be 7 or 8, score hidden from students)
– 10% Programming assignment on image compression (Median was 8, my score was in between median and 75%tile, likely to be 8 or 9 score hidden from students)
– 25% Midterm test (1 A4 sized helpsheet) (I scored 72/100, median is 69.5)
– 55% Final examination (1 A4 sized helpsheet)
– This module is about the basic mathematical theory behind signal and visual data processing. For some strange reason, this module has 70 students this year, which I guess is due to the word ‘data’ in the module name. Don’t be fooled by the pre-requisite of MA2213. The reality is that many more modules are needed to gain an appreciation of concepts taught. MA4229 Approximation theory and a module about Fourier analysis would be helpful. This module begins with a refresher course in what a basis is, with an extension to infinite dimenional space. Fourier transform is then introduced, and everything builds upon that. Discrete Fourier transforms, convolutions would be introduced, followed by the groundbreaking Multiresolution analysis and finally Wiener filters.
– Prof Ji is extremely knowledgeable about this field. He tries his best to teach along the lecture notes, and gives suitable digressions along the way by saying some interesting things and showing us some applications of the concepts he taught in class. However, his notation can sometimes differ from the lecture notes, or some serious typos can occur in the notes or when he is writing on the board.
– Prof Ji teaches the tutorials personally, and his questions are not easy. The techniques require some divine intervention sometimes. His tutorial solutions can be a little brief with some jumps in logic, so attending the tutorials do help to some extent when he explains his reasoning.
Head over to our Shop for more module content!