Alfredo Canziani   @alfcnz

Musician, math lover, cook, dancer, 🏳️‍🌈, and an ass prof of Computer Science at New York University


  Tweets by Alfredo Canziani  

Alfredo Canziani    @alfcnz    11/30/2021      

What I *really* like are these “animations”. They are unbelievable instructive.
    1         9

Alfredo Canziani    @alfcnz    11/30/2021      

😍😍😍 And we all can have access to it by heading to what link? 😬😬😬
    1         1

Alfredo Canziani    @alfcnz    11/29/2021      

OMG, I have a new inspiration source! 🤩🤩🤩 Below you can admire a few slides from @ArthurGretton lecture on Probability Divergences and Generative Models. The digrams' colours match the maths. The font is amazing. The toy illustrations are solid.

Alfredo Canziani    @alfcnz    11/29/2021      

Detailed chaptered paper video tutorial with precise video editing introducing first the maths of «The Sensory Neuron as a Transformer» and translating it into code by @moverfitted. Have a look, I think it's terrific! 🔥🔥🔥
    1         3

Alfredo Canziani    @alfcnz    11/25/2021      

SAVi, Slot Attention for Video, learns object-centric representations by predicting the optical flow of objects it's been cued on. This additional weak supervision signal lets the model generalise beyond the training distribution to novel objects, backgrounds, and video length.

Alfredo Canziani    @alfcnz    11/24/2021      

Boltzmann machines are stochastic Hopfield nets with hidden units and can be used to learn the regularities of our data. The added noise allows the model to climb energy walls and land at wider lower minima. Restricted BMs allow us to speed up inference and training…
    1         8

Alfredo Canziani    @alfcnz    11/23/2021      

I love figure 3! 😍😍😍 A few keywords for future searches: PyTorch, TensorFlow, OpenCV, PIL, image rescaling, downsample, decimation, aliasing.
    1         3

Alfredo Canziani    @alfcnz    11/17/2021      

But… where do these Hopfield nets come from?? 🧐 A spin glass 🔄🍸 is characterised by a metastable magnetic 🧲 config that can be used to encode “memories” 💭. The Ising model tells us how the magnetic moments 🧭 combine to give us an energy 🔋 that admits multiple minima 📉.
    7         39