Nikhil Thakurdesai

Senior Data Scientist at Walmart | Previously Research Assistant @ IU Computer Vision Lab | Deep Learning enthusiast

[Fabrizio Romano @FabrizioRomano]

🇵🇹 Cristiano Ronaldo scores two goals on penalty for Al Nassr and makes it 5️⃣3️⃣ goals in 2023 calendar year.

Current situation 👟✨

◉ Cristiano Ronaldo — 53
◎ Harry Kane — 52
◉ Kylian Mbappé — 52
◎ Erling Haaland — 50

[Matthew Leavitt @leavittron]

It was a blast working with @RanadiveOmkar, @nikhilt_1997, @arimorcos, @StephaneDeny to dig into WHY class selectivity gets learned even tho it's not strictly necessary for high performance. Short answer: it's transiently necessary in the early phase of training. Long answer:

[Omkar Ranadive @RanadiveOmkar]

Our paper "On the special role of class-selective neurons in early training" is now accepted at TMLR!

Link: https://t.co/G4pW5Gj5zV

w/ @nikhilt_1997, @arimorcos, @leavittron, @StephaneDeny

[Nikhil Thakurdesai @nikhilt_1997]

@FabrizioRomano Tweet

[Omkar Ranadive @RanadiveOmkar]

Work done with amazing collaborators @nikhilt_1997, @arimorcos, @leavittron, @StephaneDeny

Paper: https://t.co/cqMBlHZLib

7/7

[Omkar Ranadive @RanadiveOmkar]

Are you curious about how machine learning can be applied to perform early detection of wildfires?

Come check out our work at the @ClimateChangeAI workshop at #NeurIPS2022

Paper: https://t.co/wrNVuL3kcJ

w/ @alchera_ai, @FireScoutAI

[Nikhil Thakurdesai @nikhilt_1997]

Image patches are all you need?

[Kosta Derpanis (back home, Toronto ❤️) @CSProfKGD]

Repeat after me, another day, another MLP architecture

https://t.co/XdOGML9Vmy

[Google AI @GoogleAI]

Today we present a systematic study comparing wide and deep #NeuralNetworks through the lens of their hidden representations and outputs. Learn how similarities between model layers can inform researchers about model performance and behavior at https://t.co/2iM9HBDQjC

[Jure Leskovec @jure]

By popular demand we are releasing lecture videos for Stanford CS224W Machine Learning with Graphs which focuses on graph representation learning. Two new lectures every week. Videos: https://t.co/wD286ewlzn
Syllabus: https://t.co/gkD71I61xW

[Nikhil Thakurdesai @nikhilt_1997]

@karanjoisher That's what happens when you burn your laptop playing CyberPunk all day 😂😂

[Nikhil Thakurdesai @nikhilt_1997]

Using AI for what it's meant to be 😂😂

[Kosta Derpanis (back home, Toronto ❤️) @CSProfKGD]

You won’t see these paintings the same again 🤣

[Kosta Derpanis (back home, Toronto ❤️) @CSProfKGD]

Free access to the #ECCV2020 content. This will be until May 2021 (after which the site closes).

See: https://t.co/slleuuxST9

You have to register at https://t.co/uUTivo8fnS to get access. But there is no cost - it's free.

[Nikhil Thakurdesai @nikhilt_1997]

Check this out!

[Omkar Ranadive @RanadiveOmkar]

We are excited to share the Earthquake Detective dataset - The first crowdsourced dataset on potentially triggered seismic events.

Come check out our work at the #ai4earth workshop at #NeurIPS2020

Paper: https://t.co/zUxCPjDZTi
Code: https://t.co/eqWJoBoWQb

w/ @suzmologist

[Nikhil Thakurdesai @nikhilt_1997]

Along with proper spacing between the functions.
- One line for functions inside a class
- Two lines for functions outside a class.

[Santiago @svpino]

Python 🐍 Tip

Four naming conventions to remember:

1. Classes should use CapitalName format

2. Use lowercase_underscore format for functions, variables, and attributes

3. Use _leading format for protected attributes

4. Use __double_leading format for private attributes

[Nikhil Thakurdesai @nikhilt_1997]

This is amazing! @NVIDIAAI
What happens when there are multiple people in the background? Or when the initial keyframe sent contains only the side view of the person?

[hardmaru @hardmaru]

Neural network video streaming SDK from @NVIDIAAI can compress video conference data like these at ~0.1KB / frame, roughly 1000x better than H.264 (MPEG-4) compression on the same data (~100KB / frame).

[Nikhil Thakurdesai @nikhilt_1997]

Beautiful visualization!

[Ari Seff @ari_seff]

KL divergence asymmetry

[Nikhil Thakurdesai @nikhilt_1997]

This is great!

[Nikhil Thakurdesai @nikhilt_1997]

Interesting!

[Jiaxuan You @youjiaxuan]

How does the graph structure of a neural network affect its predictive performance?

Our ICML 2020 paper "Graph Structure of Neural Networks" https://t.co/chlJX6Qlr3 reveals many interesting findings on this topic.
with Jure Leskovec, Kaiming He, Saining Xie #ICML2020