Segmentation of different types of bones using Thresholding and morphological operations and detected (located) fracture using Hough Transform in MATLAB. Tried with Deep Learning (ML) algorithms for better accuracy.

MATLAB Code for Linear & Logistic Regression, SVM, K Means and PCA, Neural Networks Learning, Multiclass Classification, Anomaly Detection and Recommender systems.

This repository is giving some implementations of ML algorithms in MATLAB

Coded XGBoost ML algorithm fro Scratch in MATLAB

Personal projects/ Pet projects built by bare bone implementation of ml and dl algorithms and models in MATLAB

Implemented a supervised ML program in Matlab which predicts with 80% accuracy whether there is a cat in a given picture and a K-Means algorithm.

GREEN STEEL project - MATLAB - Simulink ML based digital twin of Blast Furnace with Predictive Analysis. Applied Machine Learning techniques- LSTM-based time-series forecasting and nonlinear regression using the Levenberg–Marquardt (LM) algorithm to predict CO₂ emissions in a hydrogen-injected blast furnace

No description available

Various ML algorithms written in Matlab

A set of ML algorithms implemented in MATLAB

Implementation of popular algorithms from ML coursework in Matlab

applying basic ML algorithms for classification, clustering, dimensionality reduction in Matlab

A few ML algorithm implementations from scratch in MATLAB

Learning syntax for ML algorithms in Python; some MATLAB first-principles implementations.

This repository contains implementation of ML algorithms in Matlab for the Andrew Ng machine learning class

A ML Project about Comparing Machine Learning Algorithms Performance with CDC Questionnaire in Predicting Diabetes using Python & MATLAB.

This repository contains popular Machine Learning algorithms implemented from scratch in Matlab. They are the completed exercises from Andrew Ng's ML course on Coursera

This repository contains small projects based on Embedded Image Processing. All the code is mostly in MATLAB including image preprocessing and ML algorithms applied for specific tasks as Mentioned.

I will share Matlab/Python code of various ML algorithms which were implemented on datasets available in public repositories. Instead of using ML/DL packages right away, I found coding the algorithms as a helpful initial step for good understanding. Please feel free to refer. Happy learning!

Implementation of machine learning algorithms in Matlab. The aim of these programs is to understand the detailed working of the ML algorithms instead of relying on black-box implementations of the same. Many of these programs were used for completing projects and homeworks for UF course EEL 5840 - Elements of Machine Intelligence.

Comprehensive bioinformatics system in MATLAB for analyzing STRING PPI networks, identifying disease signaling pathways via Dijkstra’s algorithm, and training ML models (SVM, Random Forest) for disease prediction based on GEO gene expression data.

In this course, I learnt the fundamental concepts of Machine Learning. It includes a detailed mathematical implementation of various ML algorithms. As part of the coursework, I implemented Linear Regression, Logistic Regression, Neural Networks, Support Vector Machines, K-Mean Clustering, Anomaly Detection and Collaborative Filtering Recommendation Systems without using any existing libraries in Matlab. This helped me to get a clear understanding of the working of these algorithms and how the algorithms are efficiently implemented using vectorisation.

Build an algorithm in MATLAB using ML techniques to predict if a person is having COVID-19 or not depending on the existing medical conditions. Further research has been conducted on identifying the most suitable machine learning techniques and increase their prediction accuracy.

Here is the code of the pipeline I created to analyse the calcium imaging movies. It contains 3 main components: 1) image stabilisation 2) component extraction and 3) spike inference. In principle there is nothing complicated in this code, it just contains the combination of different methods in one script ready to be used for your tiff stack with the calcium imaging. You can find the code here. Soon I will add it to the github page. Do not download the data and test data files, they are huge. Instead you should create data folder yourself. Here are some advices and code sources for the pipeline of ca imaging analysis. First you should be careful with the data size, your RAM will become very busy pretty quickly. Therefore if you have a large data size, you should better split it into spatial patches and process independently (better after image stabilisation) or split it into the temporal patches and process independently as well. In the first case you will have to get rid of the components because some neurons might be split between different spatial patches, while in the second case you will have to find components corresponding to the same neurons (or locations as in case of wide field imaging data). Instruction: 0) Run cvx package setup. To do so go to ./Code/cvx/cvx.setup and run it in Matlab. 1) Change the directory in Matlab where ImageCorrection_SpatialComponentExtraction_SpikeSorting.m is located (it is needed to avoid hard-coding of the packages) 2) Put your data into ./data folder in the tif stack format. 3) Make a link to the data in the script ImageCorrection_SpatialComponentExtraction_SpikeSorting.m % SET UP WORKING DIRECTORIES % input file name name = './data/840f_4xzoom16_7.tif'; - path to the tif file % output for the motion-corrected image outputFileName = './data/840f_4xzoom16_7_motion_corrected.tif'; - path to the output, it will safe only the results in the form of mat file % output for the spike-extraction savepath='./data/Cells_7/'; - folder where you want the inferred temporal components to be extracted in mat files Tips, caveats and references: 1) Image stabilisation part is taken from Eftychios Pnevmatikakis code: https://github.com/simonsfoundation/NoRMCorre By default in the script there is a rigid motion correction, which requires much less RAM. But if you want to perform the non-rigid image stabilisation, comment this section % set parameters (first try out rigid motion correction in parallel) %%{ options_rigid = NoRMCorreSetParms('d1',size(Y,1),'d2',size(Y,2),'bin_width',50,'max_shift',15,'us_fac',50); tic; [M1,shifts1,template1] = normcorre(Y,options_rigid); toc M_final=M1; and uncomment this one % do the non-rigid motion correction (longer, but better results) %{ options_nonrigid = NoRMCorreSetParms('d1',size(Y,1),'d2',size(Y,2),'grid_size',[32,32],'mot_uf',4,'bin_width',50,'max_shift',15,'max_dev',3,'us_fac',50); tic; [M2,shifts2,template2] = normcorre_batch(Y,options_nonrigid); toc M_final=M2; %} Be aware, that non-rigid motion correction might provide better results, yet it could require the substantial amount of RAM. Matlab can get crushed if it takes too much. 2) Component extraction part is done also using Eftychios Pnevmatikakis code: https://github.com/epnev/ca_source_extraction The short description of the method could be found in this presentation: https://docs.google.com/presentation/d/1ooyBtgxBSyzy3TKDVW3A-ONG8X1v34SsKei1Jz88ME8/edit#slide=id.p See the original paper here: http://www.cell.com/neuron/abstract/S0896-6273(15)01084-3 Shortly speaking the algorithm aims to find the decomposition of the calcium movie into the spatial and temporal matrices. It does that by minimisation of variance of ca imaging data. Therefore image stabilisation part is very important, since motion artefacts could provide substantial distortions for the Ca signal. Another caveat is the number of the inferred components. The algorithm is non-convex, so there are multiple solutions possible. Practically speaking it seems that the algorithm works the best when you initialise larger number of the spatial components than there are neurons. There is an algorithm for merging the components determined by merge_thr parameter. If the components look too similar, try to decrease its value. 3) The last part corresponds to the spike extraction. I used the latest algorithm of inferring spikes called ML-spike developed by Thomas Deneux et al. You could find it here: http://www.nature.com/articles/ncomms12190 Short description of this principle could be found in this movie: https://www.youtube.com/watch?v=EoWY2VeFUlM&feature=youtu.be This algorithm analyses the calcium trace and finds the spiking sequence with the highest likelihood. The precision of spike extraction strongly depends on the calcium model being used. It determines the non-linearity of the response, when multiple spikes fired at once. In the present code, the algorithm uses non-linearity specific for GCamp-6f. If the experiments are done using the other calcium sensor, you should change these parameters par.a = 0.034; par.tau = 0.76; par.pnonlin = [0.85 -0.006]; Look up the values in the supplementary of the corresponding paper, they have been calculated for GCamp-6s, OGB and some other dyes.