adding some hours that i forgot to add in other sessions... its all part of one project called logic insights..
Made a MLP from scratch! It can actually learn and predict from tabular data! It only has Dense layer for now but more type of layer to come! git repo link: github.com/LogSentry/LogSentry-Insights/tree/Confidential check it out! Note: the layer by layer analysis is not completed which is why it is not on the file but it is to come!
LogicInsights: Made a lib called net_ninja that is supposed to make more features from tabular network data processed by CIC Flow Meter git repo link: github.com/LogSentry/LogSentry-Insights/tree/test
Added an LLM to work on the theoretical output of the pipeline above it but the model it's self is an LLM so it can be used for anything! (little tweaking req ofc)... used transfer learning to teach it how to respond in the wanted format and used HuggingFace for the model... More to come in the next weeks! GitHub Link: github.com/LogSentry/LogSentry-Insights/tree/test
Working on a solution that will detect attacks or breaches based on network logs! This is first of many scrapbooks to come on this topic! For now, I made a custom script that can extract more 50+ features from PCAP files and convert it to CSV. I also made a tool for extracting from a google drive. GitHub Link: github.com/LogSentry/LogSentry-Insights/tree/test
Made 2 AI models, one to classify handwriting using MNIST, another classifying objects using CIFAR-10... GitHub link: github.com/KanishkThamman/cifar-10
made website that lets you roll a dice with physics (for when you have to roll a dice but don't have one).. git link: github.com/KanishkThamman/3D-Dice
Uses liner Regression and decision trees to predict India house prices. Used GrindSearchCV to tune one of the models. GitHub link: github.com/KanishkThamman/India-Housing
used TREE.js and othe lib to make a website that would host a 3D model of an Automata. link: kthamman.me/Automata
Finished V2 of my website! It is a dynamic website now with whole new modern design! Link: kthamman.me ( scrapbooking again since I forgot to link some sessions to the previous scrapbook)
Created a Neural Network from the MNIST dataset to classify handwriting numbers. Git link: github.com/KanishkThamman/MNIST-Neural-Net I made some changes.. Adding xdata and ydata transformations to improve AI model performance... got it to 98% accuracy
made multiple AI models using 2 different datasets.. learnt a lot while doing so and also made a very detailed README for people to learn from... Also 2 other README for challenges and things done for each dataset.. Here is the repo link: github.com/KanishkThamman/Binary-Classification-Indian-Liver-Patient-Records
