Graph Attention Network Project Walkthrough
ฝัง
- เผยแพร่เมื่อ 10 ก.ค. 2024
- ❤️ Become The AI Epiphany Patreon ❤️ ► / theaiepiphany
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
In this video, I walk you through my recently open-source GAT project. I focus on 2 potential pain points in the implementation and I also highlight some of my previous projects you could find interesting.
You'll learn about:
✔️ Cora dataset
✔️ Highly-optimized GAT implementation
✔️ Other exciting DL projects you could play with
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
✅ My GAT project: github.com/gordicaleksa/pytor...
✅ Naive neural style transfer for videos: github.com/gordicaleksa/pytor...
✅ Planetoid: github.com/kimiyoung/planetoid
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
⌚️ Timetable:
00:00 Intro to GAT project
00:35 My other deep learning projects
04:00 README walkthrough
07:35 Node degree statistics
10:10 Entropy histograms
12:05 t-SNE plots
12:50 Graph drawing layout
14:00 Jupyter walkthrough
15:22 Understanding Cora dataset
18:19 Feature vectors and labels
20:00 Building the edge index
22:40 Toy example (understanding the implementation)
29:00 Lifting
32:50 Neighborhood aware softmax and aggregate
38:18 Outro, exciting deep learning projects
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
💰 BECOME A PATREON OF THE AI EPIPHANY ❤️
If these videos, GitHub projects, and blogs help you,
consider helping me out by supporting me on Patreon!
The AI Epiphany ► / theaiepiphany
One-time donation:
www.paypal.com/paypalme/theai...
Much love! ❤️
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
💡 The AI Epiphany is a channel dedicated to simplifying the field of AI using creative visualizations and in general, a stronger focus on geometrical and visual intuition, rather than the algebraic and numerical "intuition".
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
👋 CONNECT WITH ME ON SOCIAL
LinkedIn ► / aleksagordic
Twitter ► / gordic_aleksa
Instagram ► / aiepiphany
Facebook ► / aiepiphany
👨👩👧👦 JOIN OUR DISCORD COMMUNITY:
Discord ► / discord
📢 SUBSCRIBE TO MY MONTHLY AI NEWSLETTER:
Substack ► aiepiphany.substack.com/
💻 FOLLOW ME ON GITHUB FOR COOL PROJECTS:
GitHub ► github.com/gordicaleksa
📚 FOLLOW ME ON MEDIUM:
Medium ► / gordicaleksa
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
#gat #graphattentionnetwork #project
Lol I thought the background music is louder I was using my headphones but it turns out you can't hear anything on mobile. 😂 Hopefully this will make it a bit easier to understand the project!
This is magnificent. I really liked the fact that you focused on the most difficult part of the code.
Wow amazing work! you've tried ur best to explain many details! thank you so much!!
Amazing!!! I got lot of help from your explaination in jupyter, especially some tiny concepts like cora visualization, thank god!!! (you)
Amazing! Super useful stuff!! Thanks for your clear walk-through.
Glad to hear that!
Yea dude, I can't tell you what a relief it was finding you. You're business slogan should be something like "aie, where you can actually read the scripts, run them, and start learning this shit" well... you might want something, um, catchier but I'm serious about that sentiment, keep it up
Hahah thanks duder! 😂 I love the combination of doing the engineering/coding and research otherwise there is a gap in understanding. That's my feeling, tnx again, appreciate the support!
38:18, very informative video! It took me few hours to grasp the 40mins video but it's nothing compared to reading the actual source code which may take longer hours to review it.
awesome! great explanation. Thanks
Amazing ! Great video
Thank you!
Love your content!
Thanks!
Nice work!
Thank you!
very useful!
odlično
Thank you!
How did you get dim 8 from the 3x1433 to 3x8? Excellent work...simplicity is a state of the art... :)) cheers
Great content, thank you. Do you have anything on link prediction using graph network.. Plz do share
I am planning to use GNN on a Bio knowledge graph but unable to find a suitable dataset. Can you recommend any dataset? Where can we find a protein-protein dataset?
Sorry, nothing in O(1) I'd have to investigate it myself
Can you also do some project for inductive learning tasks? Something related with NLP will be perfect.
Hey, can you also make a video on Graph Autoencoder, I feel that is one another branch which helps a lot in case of bipartite graphs
Amazing work
What's your youtube setup?
- The camera
- Microphone
- Software you use in recording your screen
Thank you in advance.
- cam: Sony ZV 1
- mic: Rode NTG
- rec: OBS
- prod: Da Vinci Resolve
You're welcome!