Posts

It is a necessity that anyone should follow some podcasts and blogs for his/her field. ML is not an exception. This is the stuff I find useful, and I try to read or watch them every now and then. ML People (Blogs) Andrej Karpathy Andrej Karpathy - Andrej Karpathy blog Yann LeCun Yann LeCun’s Home Page Andrew Ng https://www.andrewng.org Chris Olah Home - colah’s blog Alfredo Canziani atcold.github.io Cassie Kozyrkov Demis Hassabis Geoffrey Hinton Yoshua Bengio Kate Crawford - Atlas of AI Lex Fridman Soumith Chintala Yoshua Bengio - ruder....

This is one of my tasks which I’ve done when applying to KU-Leuven University at a position (it was later filled by someone else). At that time, I thought getting a Ph.D. is my type. It was later that I’ve understood that I have no interest in following academic pathway and becoming a researcher. But, since I have taken some time (a couple of days) to prepare this research proposal, I think publishing it wouldn’t be a bad idea....

In this explanation, I’ll be discussing methods for achieving interpretability in YOLOv5, a popular object detection model. While there have been several attempts to make YOLOv5 interpretable, it’s worth noting that none of these methods are currently compatible with the latest version of the model. My work is the marriage between PyTorch-GradCAM and YOLOv5. GitHub - jacobgil/pytorch-grad-cam: Advanced AI Explainability for computer vision GitHub - ultralytics/yolov5: YOLOv5 🚀 in PyTorch > ONNX > CoreML > TFLite Colab Demo To see how to use this module, maybe the best way is to use Colab tutorial I’ve provided here....

Yannic Kilcher Yannic is actually one of the brightest people I’ve ever met. He is knowledgeable mostly in NLP and also knows Computer Vision and some Reinforcement Learning. Here’s a list of videos from him which I liked a lot: How I Read a Paper: Facebook’s DETR (Video Tutorial) - YouTube: I recommend this one a lot. I’ve learned a lot from it. Machine Learning PhD Survival Guide 2021 - Advice on Topic Selection, Papers, Conferences & more - YouTube An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (Paper Explained) - YouTube MLP-Mixer: An all-MLP Architecture for Vision (Machine Learning Research Paper Explained) - YouTube AI Coffee Break with Letitia She is very good at NLP and Transformers....

Introduction I’ve tried many tools and tutorials to sync my obsidian notes between android and windows. They all have the same problems: Some need subscription service (Obsidian Sync) which I’m not willing to pay Some can’t handle editing the same file (Sync via GoogleDrive, Dropbox and …) Others need complicated setup and maintenance. My Solution This solution takes some time (at most 10 minutes) to setup. But, once it is finished, it works perfectly....

While there are many resources available on optimizing PC performance, I have found that some of them focus on less important or even incorrect details. I say the things that have actually worked for me. The credit for this post mostly belongs to Chris Titus Tech, as I sourced much of the information from his YouTube channel. However, I wanted to compile the key points in one place for easy reference and to ensure that I wouldn’t forget the important information....

Here I want to guide you toward setting a Jekyll website with al-folio template. This is the template I am using for my website and I love it. I have also made some contributions to make it more usable (specially on docker side). You do not need to install Jekyll, Docker, Ruby or even clone the repo in order to build and publish a website. You can do it all online and not even bother installing any of them....

Well, you have got a server (most probably linux) which you want to run your tasks on it. How should you do that and What technologies you should adopt? Windows Users This tutorial is written with bash (or zsh) in mind. So, If you are a mac or linux user, your default shell would be enough and you can go ahead with the rest of tutorial. However, Windows users need to follow this tutorial to install WSL and then Ubuntu to get a good bash....

These are primarily bookmarks I have in my browser. I thought they could be helpful to others, so I have compiled them here. For Researchers Connected Papers This is an excellent tool for researchers, especially when searching for review articles. Enter the paper title, and you’ll be able to see a beautiful graph of connected papers, as the name suggests! Research Rabbit I have always wanted a tool to find related papers to the ones I am already reading....

Introduction Many articles like this in the towardsdatascience tell us how to use Bayesian Linear Regression and why. I am not going to repeat that. I want to derive the formulas from scratch entirely. I also understand that comprehending this stuff is slightly challenging if you don’t have a strong mathematics background. Note: All of my notation is from Bishop’s book. The main difference between Bayesian and Non-Bayesian regression is that Bayesian regression assumes that weights are Random variables....