🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

most common neural net mistakes: 1) you didn't try to overfit a single batch first. 2) you forgot to toggle train/eval mode for the net. 3) you forgot to .zero_grad() (in pytorch) before .backward(). 4) you passed softmaxed outputs to a loss that expects raw logits. ; others? :)

Style transfer without a specific style image. I minimize only the content loss from neural style transfer and directly optimize brushstrokes via backpropagation. In this case, "style" is an intrinsic property defined entirely by the artistic medium.

.Nigam Shah recently published work using #AI to predict the likelihood of near-term patient mortality. On The Future of Everything podcast, he and host Russ Altman talk about the many opportunities for AI to reshape medicine. stanford.io/2IEsP2X

Nice approach combining ideas from world models with style transfer! arxiv.org/abs/1904.08410

Inman is always thought-provoking.

RIP to my friend, colleague, and AI visionary Nils Nilsson. Your work on the A* algorithm has improved countless lives (this is how we find the shortest path from A to B). I will always remember your work, but even more importantly your kindness. ai.stanford.edu/~nilsson/

New blog post: "A Recipe for Training Neural Networks" karpathy.github.io/2019/04/25/rec… a collection of attempted advice for training neural nets with a focus on how to structure that process over time

Exploring And Attacking Neural Networks With Activation Atlases - youtube.com/watch?v=XSWqLb…

Our new paper, “Reinforcement Learning, Fast and Slow”, reviews recent techniques in deep RL that narrow the gap in learning speed between humans and agents, & demonstrate an interplay between fast and slow learning w/ parallels in animal/human cognition: cell.com/trends/cogniti…

Quite enjoyed “A Crack in Creation” by Doudna & Sternberg - reads a bit like Watsons’s The Double Helix combining story and science, CRISPR. Not dumbed down, good discussion of the substantial repercussions (animal/plant/human soma/germ line DNA editing, gene drives, etc).

Beautiful explanation of transformer neural networks jalammar.github.io/illustrated-tr…

Great insights on learning from Leon Bottou

arxiv.org/abs/1905.02175: a strange imagenet-like dataset with very wrong-looking labels, yet a model trained on it does totally well on the normal validation set. It's a crime against ML!

There is an ongoing misconception that AI/ML are intrinsically valuable, and that therefore working in the field is bound to make you rich. A ML model is only as valuable as the problem it solves. ML without an application isn't worth anything (beyond intellectual curiosity).

the Mona Lisa finally speaks to us!! great paper: arxiv.org/abs/1905.08233… #AI living #portraits

I can hold two contradictory ideas in my head at the same time either as a thought exercise or when I genuinely don't know which is more correct. I think this is essential for learning and innovation. Doubt is fuel for discovery. Or perhaps I'm wrong in thinking this.

東京新型横断歩道。 Tokyo New crosswalk style. #モーショングラフィックス #tokyo #motiongraphics #motiondesign #cinema4d #c4d #aftereffects #fantasy

Human movement is mesmerizing. Moving elegantly in physical space is extremely difficult to re-create, but even simulation of such movement is hard. Shown is amazing work by Thomas Geijtenbeek that incorporates a model of muscles, neural delay, and biomechanical constraints:

Glad to announce that our paper on XAI for NILM has been accepted at ICASSP 2023 (IEEE ICASSP)! Check it out: pureportal.strath.ac.uk/en/publication… EEE Strathclyde (Lina Stankovic, VladStankovic) and UnivPM (Giulia Tanoni, Emanuele Principi) GECKO_ITN

📢 Excited to announce that our recent work has been published in the IEEE Transactions on Consumer Electronics! 🔍 Topic: Enhancing Transparency in Load Disaggregation using Explainable AI Link to paper here: ieeexplore.ieee.org/document/10198… VladStankovic Lina Stankovic GECKO_ITN