XOutlier
Download .zipAdd to Chrome

Alfredo Canziani

@alfcnz146k followers

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

Open on X
Top outlier
281×
Best organic post
Peak views
535k
Single post
Bookmarks
22k
Across 96 organic posts
Last scan
Oct 16
Community cache
Posts 96 posts
#DatePost↓ OutlierViewsLikesRepostsBookmarks
1May 18, '21🤣🤣🤣281×0106k22k1.1k
2Nov 19, '25Wild how a clown who has never trained a neural net feels entitled to criticise an expert who dedicated his life to pushing this field forward while also teaching all the non-obvio19.2×535k1.6k55219
3Aug 24, '23«Chapter 6» In this chapter, we'll introduce several geometric structures, over which functions are defined, and whose properties can be exploited to reduce computations and ease l14×389k1.5k168961
4Jan 23, '22«Introduction to Probability for Data Science» by @stanley_h_chan, with video lectures, exercises, and Python, MATLAB, Julia, R code! 🤩🤩🤩7.9×02k365813
5Jan 1, '23Let's start the year by brushing up on the basics of neural nets: linear and non-linear transformations. In this episode, we're concerned with inference only. Forward and backwards6.0×168k1.2k180520
6May 10, '20«A 2020 Vision of Linear Algebra» six brief videos for teaching and learning linear algebra by… yasss! Him! A 85 years old William Gilbert Strang! ❤️❤️❤️ I hope to have just a frac5.7×01.5k297502
7Aug 14, '20Who on Earth is this?! 😍😍😍 Steven Brunton from @UofWA! This is absolutely nuts! 🤯🤩🥳 Laplace transform video: Check him out! The bar keeps rising! I'm in awe 😍 To infinity an5.5×01.5k252462
8Dec 31, '23Ups and downs, downs and ups. It’s the last day of the year, I have no job offer for the upcoming one, and therefore no possibility to get a visa. Been in love with someone who end5.4×151k459352
9Jan 18, '23I taught 4 hours of Deep Learning to a class of 7th graders. I didn’t dumb it down at all. I just used the same analogies and explanations I use with the grown ups. By the end I wa5.2×146k95657156
10Sep 19, '22In «Deep Learning with PyTorch Step-by-Step» @dvgodoy presents a structured, incremental, and from first principles approach to learning PyTorch with beautiful diagrams as if he is5.2×01.1k189680
11Sep 19, '24«Linear Algebra and Its Applications» by David & Steven Lay, and Judi McDonald is truly a masterpiece. A textbook teaching Linear Algebra geometrically, application first, with5.0×140k2.2k2702.3k
12Dec 28, '22Getting emails from the other side of the world keeps me motivated in doing what I’m doing. I’m making the difference not by publishing but by educating on a global scale. I strive4.6×129k60822142
13Jan 28, '20💥 Starting the 2020 edition of @ylecun's Deep Learning class with ~200 students! 🤩 This year we *will* end up with annotated ✏️ video recordings 🎥 and publishable lecture notes 4.3×01.3k243255
14Jun 9, '25Releasing the Energy-Book 🔋 from its first appendix's chapter, where I explain how I create my figures. 🎨 Feel free to report errors via the issues' tracker, contribute to the ex4.3×120k76792641
15Nov 9, '21Wow! Exciting thread of yet another free course on Math for ML! 🥳🥳🥳 ✖️➕➖➗ «Math4ML: in the class, we cover core ideas from linear algebra, calculus, and probability that are use4.0×0895165476
16Dec 13, '24Convolutional Differentiable Logic Gate Networks @FHKPetersen3.8×105k1.6k123683
17Jul 26, '22As per your request, the latest updates on the book.3.5×0813106444
18Jan 31, '20On Tuesday, in my class, we have learnt that all a neural net does is stretching / contracting the space fabric. For example this 3-layer net (1 hidden layer of 100 positive neuron3.5×01.1k207170
19Jan 2, '24On my way to the US Embassy in Milan I got my NYU CS & DS double affiliation job offer. Right on time 😌 Alright, one step at the time let’s keep move forward. 🤓 After that I 3.4×95k37328
20Apr 3, '21«Rich» is a Python library for rich text and beautiful formatting in the terminal. Rich makes it easy to add colour and style to terminal output. It can also render pretty tables, 3.3×01.1k149212
21Jul 6, '20Week ⑪ videos 🎥 are up! 🎉 Lecture: Practicum: Transcript: Learn about: • all @PyTorch activation and loss functions; • how to make predictions and training policies under uncerta3.2×0796215273
23Apr 29, '20🥳 NEW LECTURE 🥳 Graph Convolutional Networks… from attention! In attention 𝒂 is computed with a [soft]argmax over scores. In GCNs 𝒂 is simply given, and it's called "adjacency 3.2×0873202237
24Mar 10, '21Since it seems you've been liking these new special effects 🎆🎇 quite a bit, here you have a new episode of «Foundations of Deep Learning», where I introduce vec2seq, seq2vec, seq3.2×0962125233
25Mar 24, '23Done with the VAE chapter! 🥳🥳🥳 Two sections to go and the first draft ships! 🥳🥳🥳 Yay! 🥳🥳🥳3.2×89k5723187
26Jul 19, '22A neural net is a sandwich 🥪 of linear and non-linear layers. Last week we've learnt about the geometric interpretation of linear transformations, and now we're appreciating a few3.1×0993100251
27Oct 18, '21Let's talk about maximum likelihood loss (slide 3) and its gradient wrt the parameters (slide 4). ℓ[y, F(x, 𝒴)] = F(x, y) + 1/β log ∑_y' exp[−βF(x, y')] = F(x, y) + softminᵦ[F(x, 3.0×0853132249
28Aug 30, '20Week ⑬ videos 🎥 are up! 🎉 Lecture: Practicum: Transcript: Learn about: • conv on graph as template matching; • … using spectral 👻 theorem; • (an)isotropic GCNs; • implementing g2.9×0704168261
29Dec 1, '22An awesome @geoffreyhinton presenting «The Forward-Forward Algorithm for Training Deep Neural Networks». A training algorithm that does not require to propagate derivatives or reme2.7×0790104248
30May 13, '21Another math-magician from YouTube! 😍😍😍 In this episode, @TreforBazett beautifully explains how to solve constrained optimization problems using Lagrange multipliers. This video2.7×0734105273
31Sep 18, '19My 13yo middle-schooler explaining to a bunch of PhD students her summer project: estimation of PM2.5 🏭 from daily images 📸 of Beijing, Shanghai, and Phoenix. She compared a clas2.7×01.1k12338
32Dec 15, '23LOL 🤣🤣🤣 #NeurIPS232.5×69k5603538
33May 3, '25Here we go, another post without attribution. This time we’re stealing from my friend Gregory Gundersen. I highly recommend checking out his blog at2.4×67k49729619
34Jun 12, '24Hi Twitter! 👋🏻👋🏻👋🏻 I'm back online with a new undergraduate course on Artificial Intelligence! 🥳🥳🥳 YouTube playlist: Course website:2.3×65k53077377
35Sep 28, '20Week ⑭ videos 🎥 are up! 🎉 Lecture: Practicum: Transcript: Learn about: • structured prediction with energy-based factor graph; • graph transformer net; • Viterbi algo; • Lagrangi2.3×0613133192
36Nov 29, '21OMG, I have a new inspiration source! 🤩🤩🤩 Below you can ✨ admire ✨ a few slides from @ArthurGretton lecture on Probability Divergences and Generative Models. The digrams' colour2.3×060989226
37Jul 17, '20🤓 New lectures! 🤓 Going from math and notebooks to training on 1 or more GPUs, TPUs, nodes, automatic logging through web apps, transparent handling of Slurm, and more. In short,2.2×08444798
38Jun 2, '22⚠️ Long post warning ⚠️ 5 years ago, for my birthday, out of the blue (this was so much a prank) *The Yann LeCun* texted me (no, we didn't know each other) on Messenger offering me2.2×086725114
39Nov 23, '24Random student shows up on Friday evening: prof, I didn’t get xyz. Can you explain again? Me: it’s Friday evening. Student: *puppy eyes* Me: okay, okay, let me fetch some coloured 2.1×60k93227168
40Jun 9, '20Week ⑨ videos 🎥 are up! 🎉 Lecture: Practicum: Transcript: Covering group sparsity, world models, energy 🔋 GANs! Getting used to remote teaching and escaping my tiny square! Also2.1×0551103206
41Jul 14, '21📣 NYU Deep Learning SP21 📣 Theme 1 / 3: Introduction Video lectures, notebooks, slides, and HOMEWORK have been released! 🥳🥳🥳 Website: Lecture 1: Lecture 2:2.1×0518105200
42Apr 11, '24Every time I encounter precision and recall I need to brush up their definitions for forgetting them the next day 😭. If that’s true for you as well, say no more 🥳! In today’s lec2.1×58k65781482
43May 27, '20«End-to-End Object Detection with Transformers» reasons about the relations of the objects and the global image context to directly output the final set of predictions in parallel.2.0×0576138126
44Aug 12, '21📣 NYU Deep Learning SP21 📣 Theme 4 / 3: EBMs, advanced Website: Lecture 7: Lecture 8: Lecture 9:2.0×052187207
45Jun 8, '22Another update from the book. 📖 A classifier 'moves' points around such that they can be separated by the output linear decision boundaries. Usually one looks at how the net warps2.0×066172126
46Mar 22, '22One giant leap for Alf, one small step forward for the book 🥲🥲🥲 #TeXLaTeX #EnergyBasedModel #DLbook2.0×071858101
47Jun 24, '20Week ⑩ videos 🎥 are up! 🎉 Lecture: Practicum: Transcript: Learn about: • self-supervised learning (SSL), such as PIRL with Ishan; • training a controller to drive a trailer truck1.9×0505105166
48Feb 12, '21Okay, I'm back! 😎 In this lecture we cover a few tools (@Typora, @NotionHQ, @drawio, @PyTorch, @ProjectJupyter) and all that's necessary to train a neural let to perform classific1.9×055169162
49Oct 27, '21Why to pretrain, by @kchonyc! 🤩🤩🤩 • SGD can't move very far from where it starts. • The travelling dist depends on the nb of labels. • Due to its stochasticity, it escapes narro1.9×053570174
50Jul 21, '21📣 NYU Deep Learning SP21 📣 Theme 2 / 3: Parameters sharing Video lectures, whiteboards, notebooks, slides, and HOMEWORK have been released! 🥳🥳🥳 Website: Lecture 3: Lecture 4:1.8×052889146
51Oct 31, '20This week we went through the second part of my lecture on latent variable 👻 energy 🔋 based models. 🤓 We've warmed up a little the temperature 🌡, moving from the freezing 🥶 ze1.8×056374114
52Mar 31, '19The followings deserve your attention, trust me. When I saw both I was like “Holy 🐄!”! Jupiter notebook on neural ODE (and useful links!!): Blog post on neural ODE:1.7×048214290
53Mar 26, '25While I was away, teaching for @Khipu_AI, I got ‘someone’ to teach my blackboard undergrad course. It turns out teaching (undergrad) is like riding a bike. Even though you're out o1.7×48k3902342
54Oct 10, '24Last week I gave a lecture. No one understood anything, me included. This week I gave it again, but blackboard mode. Everyone seemed to make sense, me included. 😅😅😅1.7×46k5641889
56Apr 5, '21The first episode (of five) of the energy 🔋 saga is out! 🤩 We learn about energy E and (zero temperature limit — ZTL) free energy F functions for a Latent Variable (LV) Energy Ba1.6×044570135
57Mar 28, '22Steven is a guarantee. 🤓🤓🤓 Binge-watch all his content and start with the following! Knowledge transmitted the *proper* way. What a time to be alive! 🥳🥳🥳1.6×045659146
58Aug 18, '21📣 NYU Deep Learning SP21 📣 Themes 5–8 (associative memories, graphs, control, optimisation) are online. Learn about the Lagrangian perspective of backprop! 🤩🤩🤩 Website: Lectur1.6×040882139
59Sep 26, '22Backpropagation ⏮ of the gradOutput throughout each network's module allows us to compute the rate of change of the loss 📈 wrt the model's parameter. To inspect 🧐 its value we ca1.5×043562141
60Jun 28, '21Learn all about self-supervised learning for vision with @imisra_! In this lecture, Ishan covers pretext invariant rep learning (PIRL), swapping assign. of views (SwAV), audiovisua1.5×038384143
61Jun 30, '25Currently, writing chapter 10, «Planning and control». Physical constrains for the evolution of the state (e.g. pure rotation of the wheels) are encoded through the velocity of the1.5×42k69757394
62Jul 17, '25My @NYUDataScience colleague, Carlos Fernandez-Granda, released the 700-page textbook «Probability and Statistics for Data Science», where he condenses 10 years of teaching experie1.5×42k602123686
63Mar 8, '21One more video on ConvNets is out. 😛 This time around I've sprinkled a little physics here and there, around the usual math. Yup, some special effects are included. Let me know if1.5×045651116
64Oct 19, '20Okay, let me leak something, so you can have a taste of what is coming! 😋 Of course, made with @matplotlib, @numpy_team, and @ProjectJupyter. #opensource #DeepLearning1.5×05597343
65Sep 21, '21*Finishes teaching 1.5 hours* *Students ask questions for 30 minutes* *New slides address students' doubts* Let's learn about environment setup, @PyTorch 5-step training, and why g1.5×046946113
66May 12, '20My GCN lecture is now available (in English only, at the moment) here: The video will follow, in a month or so.1.4×04399385
67Apr 22, '20🥳 NEW LECTURE 🥳 “Set to set” and “set to vector” mappings using self/cross hard/soft attention. We combined a (two) attention module(s) with a (two) k=1 1D convolution to get a t1.4×039476115
68May 15, '20Week ⑦ videos 🎥 are up, finally! 🎉 Lecture: Practicum: Transcript: Finally, we got to start the Energy Based Models section! Learn anything about latent variables and autoencoder1.4×038287108
69Dec 1, '21Let's say we want to maximise f(x, y) while having g(x, y) = g₀. At the maximum (xₘ, yₘ) location the gradients of f and g will be proportional to each other: ∇f = λ ∇g. The propor1.4×045553111
70Mar 1, '22Insane!!! 🥳🥳🥳 Kevin's books drafts publicly released and figures' code open sourced!1.4×047352102
71Jul 7, '22What does SVD (singular value decomposition) do to a matrix A? It finds a pair of orthogonal axes V that stays orthogonal U after the linear transformation A and are individually s1.4×037154140
72Oct 12, '21It's 02:00 and I'm not sleeping. Why? Because I've again decided to draw new content for class. Why? Idk, it's addicting… send help… 😭😭😭 Recap on CE and negative logits free ene1.4×04954083
73Sep 4, '25Success! Lecture 2 was recorded! 🥳🥳🥳 Maybe I got this! 😅 So here we go, with the second edition of «Intro to Deep Learning», from McCulloch & Pitts’ binary neuron to the tr1.4×38k52919221
74Sep 18, '20این سمستر من و Yann در دانشگاه نیویارک 🥳 Deep Learning را🔥 تدریس میکنیم. نوت ها و ویدیو های درسی را به شکل آنلاین قرار دادیم. به چند داوطلب نیاز داریم که ما را در ترجمه فارسی یار1.4×04159352
75Aug 19, '21So, I've been toying around with the eigen-stuff. Why? Because it pops up everywhere but I couldn't *see* how it worked. So, I took the time to learn JavaScript, @ProcessingOrg, an1.3×038854113
76Sep 14, '19My neural net minimising its loss 😁1.3×05438212
77Nov 7, '24Daily TikZ show off. 😁😁😁1.3×37k68934200
78Jul 22, '22Chapter 1 (2 and 3) completed! 🥳🥳🥳 We've seen a linear and a bunch of non-linear transformations. But what can a stack of linear and non-linear layers do? Here we have two fully1.3×04485675
79Aug 23, '21Well, since you were interested, now I have a blog. In this way, you don't have to dig my Twitter account for finding my micro-lectures. Also, I can use colours and LaTeX, so I thi1.3×036847116
80Sep 16, '21Yesterday, in @kchonyc's NLP class, we've learnt about the input (word and sentence) and class embeddings, and how these are updated using the gradient of the log-probability of th1.3×035444120
81Apr 8, '20«We install Vim on the door so that it was impossible to exit during quarantine» 🤣🤣🤣1.3×0507747
82Sep 23, '25The directional derivative of a scalar field ℒ:ℝᴺ → ℝ at w⁰ along direction û is given by: Dᵤℒ(w⁰) = ∇ℒ(w⁰)ᵀ û = ‖∇ℒ(w⁰)‖ cos ϕ, where ϕ is the angle between ∇ℒ(w⁰) and û. The dire1.2×35k58048185
83Oct 2, '20The Bellman equation is simply a system of linear equations you can solve to get your value function by following π from s. This statement was confusing, so I coded it up, and… aww1.2×03895878
84Dec 1, '22Four more posters which clearly communicate something. There are 1k posters and 2 hours of time. Plz, communicate effectively. #NeurIPS221.2×038635101
85Dec 9, '22Maybe I went a little overboard with this loaning thingy 😅😅😅 Anyhow, I really love the top 3! 🤓🤓🤓1.2×035624123
86Apr 17, '19Today we've learnt about the bubble-of-bubbles interpretation of a variational autoencoder (VAE). Its loss is the sum of the reconstruction loss and the KL divergence with a Normal1.2×03836966
87Jan 16, '22Uncle Alf teaching Python 🐍🐍🐍 programming to nephew. import turtle 🐢🐢🐢 ninja = turtle.Turtle() ninja.forward(100)1.2×05192124
88Aug 22, '20You *literally* cannot believe the EXCITEMENT 🤩🤩🤩 I have for my pupil's first video about visually explaining Deep Learning! 💥💥💥 Notation and colour scheme compatible with th1.1×04354949
89Feb 12, '2218 hrs writing the book in a row… Let's go home 😝😝😝 Good night World 😴😴😴1.1×0485947
90Oct 6, '20Today iPad lecture! Perhaps I should turn these into slides, hehe 😁 Anyhow, they are still pretty, I think 😋 I like to emphasise how Ax can be seen as template matching of A's ro1.1×04114157
92Jun 10, '22When looking at a classifier, we can consider its energy as being the cross-entropy or its negative linear output (often called logits). The energy of a well-trained model will be 1.1×03513781
93Sep 30, '21Since I think you like spinning stuff, here's the rendering of the last figure. Rigorously made with @matplotlib and @PyTorch.1.1×03724652
94Dec 20, '21New lecture in the making: «Energy-Based Self-Supervised Deep Learning — A gentle introduction». The lecture will take place Wed 22 Dec, at 14:00, at @UniTrieste, aula Morin, edifi1.0×03644052
95Apr 4, '25Professor, we came asking a few questions! […] 4 hours later (it’s 8 pm) walls of whiteboard have been filled with maths 🥹🥹🥹 @NYUDataScience @NYU_Courant1.0×28k43722104
96Feb 9, '22I think I've just acquired the title of TikZ-ninja.1.0×03781948
97Dec 2, '22OMG! 🤯🤯🤯 @soumithchintala announcing @PyTorch 2.0! Fully backward compatible. Speeds up execution without code change. Can further improve perf by issuing a single command. #PyT0.9×03924513
98Jul 2, '25To compute the movement of the state x(t), we need to temporally integrate its velocity field ẋ(t). 🤓 The control signal steering angle stays at 0, then 0.05π, then linearly to −00.8×23k37446213
99Nov 27, '21Yesterday, while sitting on the chair outside my barber in Chinatown, three Spanish speaking guys approached me. One comes forward, smiles with his eyes, and says «Thank you! ❤️». 0.8×040663

Scan another creator

@
@karpathy@levelsio@paulg@naval@marclou@illyism

Run this on any account.

XOutlier builds this table for every public profile you open on x.com, with live data and every post, not just the top 50.

Add to Chrome Download .zip