Vartan Polodyan

[Tom Yeh @ProfTomYeh]

The past two weeks were too busy to keep up with editing and uploading my lecture videos. I know many of you have been waiting to see how I teach auto-encoder and backpropagation by hand ✍️ in Excel! Here it is!

It took me a few years to invent this method to show both forward and backward passes for a non-trivial case of a multi-layer perceptron over a batch of inputs, plus gradient descents over multiple epochs, while being able to hand calculate each step and code in Excel at the same time.

With the release of this video, I am glad I finally get to share this method with the whole world! 🙌 Let me know what you think in the comment!

= Chapters =
• Encoder & Decoder (00:00)
• Equation (10:09)
• 4-2-4 AutoEncoder (16:38)
• 6-4-2-4-6 AutoEncoder (18:39)
• L2 Loss (20:49)
• L2 Loss Gradient (27:31)
• Backpropagation (30:12)
• Implement Backpropagation (39:00)
• Gradient Descent (44:30)
• Summary (51:39)

= Semester Schedule =

1. Image
2. Function
3. Transform
4. Feature
Special Public Lecture: DeepSeek
5. Compare
6. Filter 1D
7. Filter 2D
8. NN
9. MLP
10. AutoEncoder 👈 NEW
11. GAN
12. CNN
13. ResNet
14. YOLO
15. U-Net
16. Camera
17. Epipolar
18. RANSAC
19. Fourier
20. SVM
21. Probability
22. RNN
23. Transformer
24. ViT
25. CLIP
26. Diffusion
27. Latent Diffusion

---

How can you support AI by Hand ✍️?

1. Check out our store: https://t.co/aX9M0pSjlH
2. Repost ♻️

[Tim Sweeney @TimSweeneyEpic]

Simon Peyton Jones is joining Epic! Simon is a force of nature behind the Haskell programming language and has written amazing papers on type systems, transactional memory, and other fields that will be key in realizing the metaverse.

https://t.co/H1F0TQPwvg

[Satoshi Matsuoka @ProfMatsuoka]

SKA (Square Kilometer Array) Telescope keynote at #SC17. Of the HPC and Data requirements presented, FLOPS would not be hard at 250PF, but bandwidth is immense—200PB/s(!). Even the upcoming Summit has only 1/10th the memory BW. Here again BYTES the challenge in HPC, not FLOPS.

[witeken @witeken]

This, people, is what a 10nm transistor looks like. About 10 billion of those go into a square centimeter -- 2x other's "10nm".

[Geoff Teehan @gt]

If nutrition labels were designed to visualize ingredients like this, we would make better choices—as would the manufacturers.

[Louis Opter @1opter]

How the f. can people think/say that women can't write code, this is straight up assembly.

[Tim Peake @astro_timpeake]

Can we see other planets from space? Here’s Venus rising just before the Sun

[Дмитрий Стешин @D_Steshin]

Было-Стало:

[Game Developer @gamedevdotcom]

Microsoft demos 'holoportation' 3D presence tech with HoloLens https://t.co/mqRCeQUte1

[Elon Musk @elonmusk]

Interesting to think of physics as a set of compression algorithms for the universe. That's basically what formulas are.

[Scott Kelly @StationCDRKelly]

Good afternoon #Egypt! Hope to see you up close some day. #YearInSpace

[Quanta Magazine @QuantaMagazine]

Is the biological world computational at its core? Leslie Valiant thinks it is: https://t.co/Wgu0OisXYj

[Scott Kelly @StationCDRKelly]

#Andes When you believe, you can move #mountains. Or move over them. #YearInSpace

[Jon Kalb @_JonKalb]

Alex Stepanov retiring. I'm off to Alex Fest (hosted by A9). Speakers include @stroustrup , Dave Musser, @pmcjones, @SeanParent, & Dan Rose.

[Tim Sweeney @TimSweeneyEpic]

It would be better to kill Java on Android than to upgrade it. Everyone writes stuff in C++ code nowadays, or at least in C++ engines.

[Kostas Anagnostou @KostasAAA]

AMD Reveals Polaris GPU Architecture: 4th Gen GCN to Arrive In Mid-2016 https://t.co/PqbPcloBYS

[BioDigital @biodigitalhuman]

Need a #NewYearsResolution? Here are 9 that'll get you feeling your best https://t.co/o4uPYSmw23 #healthynewyear

[Eric Haines @pointinpolygon]

Wait, what?! Most math texts downloadable for free! Plus some older books in other areas. Merry Christmas, Springer: https://t.co/Qqfl9IbwvL