markoharalovic/information_processing

Repository exploring signal processing, bioinformatics, and image analysis using Fourier transforms, hidden Markov models and Convolutional Neural Networks

Information Processing Course

This repository consists exploration of signal processing, bioinformatics, and image analysis using Fourier transforms and hidden Markov models. It is based on the course "Information Processing" done on college Faculty of Electrical Engineering and Computing (FER) in Zagreb,Croatia.

Course Overview

The course provides an overview of basic information processing techniques and a look into several application areas with characteristic data processing procedures. Key topics include:

• Signal Processing in Python: Introduction to time-continuous Fourier series and signal processing with applications in analyzing river water levels.
• Fourier and Wavelet Transforms: Discrete Fourier transform, time-windowed Fourier transform, and wavelet transform techniques, especially in the context of river level analysis.
• Introduction to Bioinformatics: Data types in bioinformatics and alignment algorithms. DNA sequences are represented as signals, with convolution of signals used as a measure of similarity. The MAFFT algorithm is explored for sequence similarity using fast Fourier transforms.
• Hidden Markov Models (HMM): Application of HMMs for modeling time series with discrete observations, evaluating sequence likelihoods (forward and backward algorithms), finding the most likely sequence of hidden states (Viterbi algorithm), and determining optimal HMM parameters for multiple observation sequences.
• Quantitative Finance: Introduction to financial data types and the basic properties of financial time series.
• Multivariate Data Analysis: Principal component analysis and its applications in quantitative finance.
• Image Analysis and Neural Networks: Challenges in object detection, basics of operating and training artificial neural networks, classification of input samples, and object detection using neural networks.
• Data, Signals, Systems: Exploring invariants and symmetries. Profilometry using stripe projection and the iterative closest point (ICP) registration process.