50+ days of Machine Learning

 Hello :)  Today is Day 99! A quick summary of today: covered the next part of XCS224W: ML with Graphs Instead of doing the 3rd homework which is related to createing and training a GNN, I decided to continue with the next theory part, which funny enough is called 'theory of GNNs'.  Covered topics: Expressiveness of GNNs, computational graphs, problems with GCN and GraphSAGE's mean and max-pooling, designing the most expressive GNN, the Universal Approximation Theorem, Graph Isomorphism Network (GIN) That is all for today! See you tomorrow :)

 Hello :)  Today is Day 98! A quick summary of today: completed the whole 2nd homework of XCS224W: ML with Graphs started reading Vicki Boykis' 'What are embeddings?' book About the homework The topic was: GNNs Using PyTorch Geometric, but I am afraid that is as far as I can go. We are not allowed to share info about the homeworks, but it was based on my notes from Days 94, 95 and 96.  My score: I had a quick look at assignment 3 and apparently its based on Days 94, 95 and 96 as well, so for tomorrow I will either take a break and not do 2 assignments in a row by continuing to watch the next lectures, or do it as well. (the 1st option is more likely) As for the book 'What are embeddings?' by Vicki Boykis  I had it on my list for a bit and I just wanted to read it and make take some notes (which are below). But most (if not all) the info was already covered when I covered CS224N NLP with DL : from Day 57 ~ Day 65. Still it was nice to confirm my knowledge....

 Hello :) Today is Day 97! A quick summary of today: read through and reviewed my notes on intro to GNNs  (Day 94) read through and reviewed my notes on Designing a GNN layer  (Day 95) read through and reviewed my notes on GNN training  (Day 96) started 2nd homework of the course covering the above two First I had a quick look through the 2nd homework that covers material covered in Day 94 and Day 95, and I decided that it would be best to do a proper review of the important concepts from those days Topics I went over: Designing a single layer of a GNN,  message computation,  aggregation,  GCN,  GraphSAGE,  GAT,  attention and multi-head attention in graphs,  stacking GNN layers,  the problem of over-smoothing,  shallow GNNs,  using skip connections graph augmentation,  feature augmentation,  training GNNs on a node-level, edge-level and graph-level,  pooling for graph-level tasks,  DiffPool, ...

 Hello :) Today is Day 96! A quick summary of today: covered the GNN training pipeline from Stanford's XCS224W: ML with Graphs applied for a few internships this summer + found theforage My notes from the training pipeline lectures Covered topics: graph augmentation, feature augmentation, training GNNs on a node-level, edge-level and graph-level, pooling for graph-level tasks, DiffPool, supervised, unsupervised and self-supervised learning, loss functions, evaluation metrics, splitting data Looking for opportunities this summer I am in South Korea, but I still have the right to work in the UK, so I decided to go on Bright Network to look for data analytics/science internships and saw some good opportunities. I hope that if a company shows interest they would do an online interview. In addition, I found TheForage  which provides free virtual interships. Though short, they could give some interesting perspectives into the work life of a software engineer or investment banker (th...

 Hello :) Today is Day 95! A quick summary of today: learned how a GNN layer is constructed and operates ( XCS224W: ML with Graphs ) became a fellow of the Royal Statistical Society  (I thought I already was) Lecture 2.2 GNN Design Space Covered topics: Designing a single layer of a GNN, message computation, aggregation, GCN, GraphSAGE, GAT, attention and multi-head attention in graphs, stacking GNN layers, the problem of over-smoothing, shallow GNNs, using skip connections Becoming a fellow of the Royal Statistical Society I was registered as a student in the RSS and I thought I was like a regular member, but apparently I needed to register as a proper 'Fellow'. Going forward I see there are events that they organise, and have some courses on statistics through which I can climb the pyramid Also, for a bit I tried making a discord server for study groups and discussions between students of the XCS224W course, so far noone wanted to join. I am not sure how they usually do...

 Hello :) Today is Day 94! A quick summary of today: Finished the last part of Traditional ML methods for Graphs: Link analysis page rank random walks and embeddings Did assignment CoLab 1: Learning Node Embeddings Covered first part of Module 2: Intro to GNNs Today I continued with the coverage of XCS224W: ML with Graphs My notes for Link analysis page rank random walks and embeddings Covered topics: PageRank, Matrix Formulation, Power iteration method, Solutions to dead-ends and spider traps, Personalised PageRank, Random walk with restarts, Using Matrix factorization to express node embeddings based on random walks Assignment 1: Learning Node Embeddings  We are not allowed to share any of the code, and I really do not want to risk anything, so I will just say, similar colabs can be found on the course's main webpage . But I spend a lot of time to understand each line of code that I wrote, and how theory from Module 1 on node embeddings is applied to practice. Also, I g...