Chirag Nagpal @nagpalchirag Twitter profile

Last Seen Profiles

@Sri478241177294

@Routheream

@LunarSnow1

@sy_khd

@enazunic

@jellyfullmoon

@vickypangeranb1

@kizakiko_camp

@maluparis

@RRCryp

@marioduran

@tomlon_

@SL_Azkikr

@ChanGoogle25363

@aborealis940

@GolfCoalCreek

@TimacAgroArable

@RajK42165023

@RyozoraVT

@Buster_ESPN

@selenabiensur

@juplutoo

@JimmyCh58084618

@King_B_randon

@Stalkphish_io

@n1x11i

@stw_pdg

@yu7775006433528

@jwbartlett92

@Acacio_Wen

@hinashahidxo

@katsuya_choshi

@HKickboxing

@HeathlandDavid

@Neosurf

@thomasestimbre

Chirag Nagpal

@nagpalchirag

3 months

My kindest sympathies for the Bayesian Optimization folks

Jascha Sohl-Dickstein

@jaschasd

3 months

Have you ever done a dense grid search over neural network hyperparameters? Like a *really dense* grid search? It looks like this (!!). Blueish colors correspond to hyperparameters for which training converges, redish colors to hyperparameters for which training diverges.

279

2K

10K

12

41

632

Chirag Nagpal

@nagpalchirag

3 years

1) In view of the severe COVID crisis in India. We, the Students, Staff, Faculty and Alumni of @CarnegieMellon University are rallying support to petition the President of the University, Farnam Jahanian to:

Carnegie Mellon University Petition on COVID-19 Humanitarian Crisis in India

docs.google.com

5

36

104

Chirag Nagpal

@nagpalchirag

1 year

📣📣 I am on the job market for academic and industry positions. I am looking for opportunities to work on challenging problems that involve the use of data driven machine learning for supporting complex decision making in healthcare.

1

22

89

Chirag Nagpal

@nagpalchirag

2 years

Wow 🤩 @DukeU is hands down one of the nicest campuses I’ve visited! 🤩

4

72

Chirag Nagpal

@nagpalchirag

4 months

How to be a tenured cs academic faculty in 2024: from itertools import combinations ai_buzz_words = ['rag', 'rlhf', 'dpo', 'safety', 'diffusion', 'uncertainity'] for i, combination in enumerate(combinations(ai_buzz_words, 3)): gradStudent[i].writePaper(combination)

Mandi Zhao

@ZhaoMandi

4 months

We are just doing combinatorics on the hype words at this point

2

9

210

2

5

72

Chirag Nagpal

@nagpalchirag

5 months

Very excited to see our NeurIPS spotlight paper "Participatory Personalization in Classification" w/ @heyhaileyjoren @berkustun @kat_heller feature in Google Deepmind's 2023 year in review blog!

Demis Hassabis

@demishassabis

5 months

It's always great to look back on the year in a year-in-review blogpost with @JeffDean & James Manyika. It's been an amazingly productive year for us, doing awesome research, shipping products and advancing science - 2024 is going to be incredible!

38

145

965

1

6

53

Chirag Nagpal

@nagpalchirag

3 years

Happy to share I’ve been nominated for the @DAAD_Germany Fellowship to visit and collaborate with researchers in 🇩🇪. If you’re a German research group pursuing ‘AI in Healthcare’ research please do reach out! #ML4H

3

1

46

Chirag Nagpal

@nagpalchirag

2 years

🚨Presenting🚨 auton-survival: an open-source package for Regression, Counterfactual Estimation, Evaluation and Phenotyping Censored Time-to-Events MLHC '22 Paper: Repo: Docs: w/ @WPotosnak @AutonLab (1/n)

GitHub - autonlab/auton-survival: Auton Survival - an open source package for Regression, Counter...

Auton Survival - an open source package for Regression, Counterfactual Estimation, Evaluation and Phenotyping with Censored Time-to-Events - GitHub - autonlab/auton-survival: Auton Survival - an...

github.com

1

12

38

Chirag Nagpal

@nagpalchirag

4 years

Practical use of Deep Learning for Heterogenous Treatment Effect estimation is challenging owing to a lack of interpretability. At @CHILconference we propose a graphical model that allows learning interpretable subgroups while adjusting for confounders through deep learning.

1

9

38

Chirag Nagpal

@nagpalchirag

4 years

We open sourced Deep Survival Machines! Also if you haven't already checked it out, the full paper is now on arXiv

Deep Survival Machines: Fully Parametric Survival Regression and...

We describe a new approach to estimating relative risks in time-to-event prediction problems with censored data in a fully parametric manner. Our approach does not require making strong...

arxiv.org

Chirag Nagpal

@nagpalchirag

4 years

If you missed our Spotlight Talk earlier this morning at #ML4H , #NeurIPS2019 , check out our poster now! work with Xinyu Li, Artur Dubrawski, @AutonLab

3

8

27

1

10

35

Chirag Nagpal

@nagpalchirag

5 years

We arXived our paper () with State of the Art results on using Mixtures of Experts 4 Survival Anlysis. Our proposed model beats Deep Surv () on multiple std datasets, while retaining intrpretability via linearity. @mldcmu #ML4H #MLHC

DeepSurv: Personalized Treatment Recommender System Using A Cox...

Medical practitioners use survival models to explore and understand the relationships between patients' covariates (e.g. clinical and genetic features) and the effectiveness of various treatment...

arxiv.org

1

9

34

Chirag Nagpal

@nagpalchirag

5 years

@abrahamsamuel @SushmaSwaraj @mkstalin @narendramodi Clarify the meaning of denied immigration? Did he not let you enter India? Or did he transfer you to another immigration officer? The two are very different situations which might both be described somewhat misleadingly as 'denied immigration'

6

7

31

Chirag Nagpal

@nagpalchirag

2 years

1/n 📣 New Preprint: "𝑪𝒐𝒖𝒏𝒕𝒆𝒓𝒇𝒂𝒄𝒕𝒖𝒂𝒍 𝑷𝒉𝒆𝒏𝒐𝒕𝒚𝒑𝒊𝒏𝒈 𝒘𝒊𝒕𝒉 𝑪𝒆𝒏𝒔𝒐𝒓𝒆𝒅 𝑻𝒊𝒎𝒆-𝒕𝒐-𝑬𝒗𝒆𝒏𝒕𝒔" w/ @MononitoGoswami @AutonLab and @DrDufendach @UPMC #MachineLearning #EpiTwitter #MedTwitter #causalinference #DataScience

Counterfactual Phenotyping with Censored Time-to-Events

Estimation of treatment efficacy of real-world clinical interventions involves working with continuous outcomes such as time-to-death, re-hospitalization, or a composite event that may be subject...

arxiv.org

2

15

32

Chirag Nagpal

@nagpalchirag

4 years

There are enough @NeurIPSConf posters on my Toronto to Vancouver flight to have a mid air workshop! #NeurIPS2019

3

0

31

Chirag Nagpal

@nagpalchirag

4 years

No @nytimes , you're wrong. They weren't "rebels". They were 'terrorists'. Funded & trained by the same bosses and ideologies that caused to 9/11. Belonging to the same country that harbored OBL. Your double standards stem from blatant deep rooted racism .

1

8

30

Chirag Nagpal

@nagpalchirag

4 years

The self-quarantine is a good time to restudy classic Machine Learning theory. Here's @ArthurGretton wonderful slides on Hilbert spaces, RKHS and the Kernel Trick.

1

3

30

Chirag Nagpal

@nagpalchirag

4 years

The indomitable @pmichelX , @LTIatCMU with @omerlevy_ explaining if Sixteen Heads are indeed better than one at #NeurIPS2019 ! @gneubig

1

4

31

Chirag Nagpal

@nagpalchirag

4 years

In some personal news... I'll be remote interning this summer with the amazing @kat_heller and others from @GoogleHealth , @GoogleAI #MLHC #ML4H

5

0

30

Chirag Nagpal

@nagpalchirag

4 years

Unpopular Opinion: Hackathons make bad engineers; incentivize quick and dirty prototyping, instead of helping inculcate good engineering practices; and result in increasingly impatient stream of developers.

3

5

28

Chirag Nagpal

@nagpalchirag

5 years

This should be called the Bechdel Test of Academia.

Katherine Bailey

@katherinebailey

5 years

I spoke on a majority female panel today that wasn't about being a woman in tech/eng/AI 😀

4

6

227

1

2

28

Chirag Nagpal

@nagpalchirag

4 years

If you missed our Spotlight Talk earlier this morning at #ML4H , #NeurIPS2019 , check out our poster now! work with Xinyu Li, Artur Dubrawski, @AutonLab

3

8

27

Chirag Nagpal

@nagpalchirag

2 years

Had a fun weekend at @mlforhc in Durham ... met some old friends and made some new.. The conference size and niche format makes MLHC a great place to spark conversation and collaboration.. looking forward to partaking again next year !!!

0

3

26

Chirag Nagpal

@nagpalchirag

1 year

1/ Machine Learning is extensively applied to "phenotype" clinical data, BUT these models often seek answers to the WRONG CLINICAL question. For a Phenotype to be ACTIONABLE, it should discover patients that benefit (harmed) from a therapy! #ACC23

1

3

21

Chirag Nagpal

@nagpalchirag

3 years

Publicly calling out perceived mistakes in papers you’re reviewing 1) jeopardizes double blindness 2) morally objectionable; especially when the author is a young researcher. Reviewer’s job is to give cnstrctve feedbck to help advance science, not to serve as a gatekeeping bully.

1

2

21

Chirag Nagpal

@nagpalchirag

4 years

We will present the first Fully Parametric Estimator of Survival Functions at #NeuRIPS #ML4H as a Spotlight talk. Our approach allows learning non-linear representations of features in the presence of censoring and doesn't make strong assumptions of constant baseline hazard.

1

3

20

Chirag Nagpal

@nagpalchirag

3 years

From cursing Reviewer 2 to becoming the cursed Reviewer 2, I have come a long way.

2

1

18

Chirag Nagpal

@nagpalchirag

1 year

Having a fun time presenting at #ACC23 @ACCinTouch so far !

0

18

Chirag Nagpal

@nagpalchirag

2 years

“AI Governance from a control system perspective ” @krvarshney enchanting a rapt audience at @CarnegieMellon @HeinzCollege seminar!

1

18

Chirag Nagpal

@nagpalchirag

1 year

@TaliaRinger Try applying for a US visa once :)

0

17

Chirag Nagpal

@nagpalchirag

2 years

It is exciting to see our methodology, Deep Survival Machines applied in real world adverse clinical risk prediction tasks! Check out this @LancetDigitalH paper that uses DSM to predict cardio vascular risk from the UK Biobank cohort !

The Lancet Digital Health

@LancetDigitalH

2 years

NEW Research: 'Neural network-based integration of #polygenic and clinical information: development and validation of a #prediction model for 10-year risk of major adverse #cardiac events in the UK Biobank cohort'. @ChariteBerlin @HumboldtUni @UniHeidelberg

0

5

15

0

5

16

Chirag Nagpal

@nagpalchirag

10 months

@ilijabogunovic What about PhD students who are only working by themselves w/o supervision?

0

16

Chirag Nagpal

@nagpalchirag

3 years

@Iyervval @shambhav15 @ThePrintIndia @kushal_mehra @AbhishBanerj @Sanjay_Dixit @Abhina_Prakash

2

1

13

Chirag Nagpal

@nagpalchirag

2 years

Had a great time participating in the PhD student panel at the #KDD2022 Undergraduate Student consortium. Thank you for the opportunity!

0

14

Chirag Nagpal

@nagpalchirag

3 years

1) Recognize the massive humanitarian crisis in India and the stress it is causing to the Indian students amongst the CMU community.

1

5

13

Chirag Nagpal

@nagpalchirag

3 years

I’m not a fan of this whole AI that can write code thing.

3

0

14

Chirag Nagpal

@nagpalchirag

2 years

📣📣📣 Stoked to announce that our paper "Counterfactual Phenotyping with Censored Time-to-Events" would be presented as a full paper at #KDD2022 ! Paper: Code: Demo Notebook: 🧵👇 @kdd_news @AutonLab

Chirag Nagpal

@nagpalchirag

2 years

2/n Real world clinical studies involve censored time-to-event outcomes. We propose a latent variable approach to discover individuals that respond differentially to interventions, independent of base event incidence rate. We describe these as "Counterfactual Phenotypes"

1

5

1

5

14

Chirag Nagpal

@nagpalchirag

4 years

Twitter has been stressing me out, so I decided I’ll just tweet whenever I get a paper accepted and cause stress to others. #AcademicTwitter #phdlife

2

0

14

Chirag Nagpal

@nagpalchirag

3 years

I call on @PMOIndia @narendramodi to invite Indian academics in STEM to perform an independent audit of data and AI initiatives by @DataPortalIndia and @NITIAayog . Ideally involving top academics like @MilindTambe_AI @mishumausam @ravi_iitm @prfsanjeevarora

5

3

12

Chirag Nagpal

@nagpalchirag

3 years

2) Request the US administration to ease restrictions around export of critical healthcare resources, including ventilators, oxygen concentrators, raw material and intellectual property for vaccine development.

2

6

11

Chirag Nagpal

@nagpalchirag

3 years

1) Update: We received over 300 signatures on the petition from members of the @CarnegieMellon community, including over 200 students and 50 faculty members. I shall remain ever grateful to everyone who supported us. Thank you 🙏

Chirag Nagpal

@nagpalchirag

3 years

1) In view of the severe COVID crisis in India. We, the Students, Staff, Faculty and Alumni of @CarnegieMellon University are rallying support to petition the President of the University, Farnam Jahanian to:

5

36

104

2

0

12

Chirag Nagpal

@nagpalchirag

3 years

1/n New paper: 𝙍𝙚𝙘𝙪𝙧𝙧𝙚𝙣𝙩 Deep Survival Machines (with @VincentJeansel1 )- We proposed a flexible parametric approach to model time-varying covariates for Time-to-Event regression with RNNs in the presence of censoring.

MRC Biostatistics Unit

@MRC_BSU

3 years

New research published at Proceedings of AAAI Spring Symposium on Survival Prediction by PhD student @VincentJeansel1 , @SCSatCMU , @AutonLab and @nagpalchirag . 'Deep Parametric Time-to-Event Regression with Time-Varying Covariates'

0

3

8

1

3

13

Chirag Nagpal

@nagpalchirag

4 years

Personal Takeaway from #NeurIPS2019 ...( 🇮🇳 food in 🇨🇦)>>> (🇮🇳 food in 🇺🇸)

2

0

13

Chirag Nagpal

@nagpalchirag

4 years

My andrew inbox has #15213 unread mails and I can't help but appreciate how poetic this is. @CarnegieMellon #TartanProud

1

0

13

Chirag Nagpal

@nagpalchirag

4 years

Come be part of the @AutonLab . Join the illustrious ranks of our alumni including @AndrewYNg , @awmcmu and many others !

Auton Lab, Carnegie Mellon

@AutonLab

4 years

The @AutonLab at @SCSatCMU is hiring Research Analysts and Project Scientists in Full Time positions! Come be part of a leading team in Artificial Intelligence and Machine Learning! Job Descriptions below:

1

8

14

0

2

12

Chirag Nagpal

@nagpalchirag

3 years

3) Mobilizing the CMU Alumni @cmualumnihouse community with a centralized university effort to make charitable donations and support healthcare organizations in India.

1

4

11

Chirag Nagpal

@nagpalchirag

4 years

Our family dog Pepper was great at MATLAB. This is her rejoicing after having implemented a 2 layer MLP from scratch. (ca. 2004)

1

12

Chirag Nagpal

@nagpalchirag

5 years

Had great fun welcoming the incoming @SCSatCMU PhD student cohort with a Treasure Hunt and Grill! Thanks to @jessica_colnago and the rest of #SCS4ALL for the support in organizing!!

0

2

12

Chirag Nagpal

@nagpalchirag

5 years

Super impressed with @math3ma blog's on analysis. I would strongly recommend, the material is very well introduced even if you don't have rigorous mathematical training. If only @Wikipedia was more like this.

Fatou's Lemma

Fatou's Lemma, the Monotone Convergence Theorem, and the Dominated Convergence Theorem are three major results in the theory of Lebesgue integration which, when given a sequence of functions $\{f_n...

www.math3ma.com

1

0

11

Chirag Nagpal

@nagpalchirag

3 years

5) If you are a student at CMU or another university in the US and would like to help and support this cause please reach out to me via Twitter or at chiragn at cs dot cmu dot edu.

1

4

11

Chirag Nagpal

@nagpalchirag

2 years

Looking forward to present "Counterfactual Phenotyping with Censored Time-to-Events" at #KDD2022 Oral Presentation: Thursday, August 18, 10:00 AM-12:00 PM, Room 202B Poster: Monday, August 15, 7:00PM-8:30PM If you're around, come say hi! Happy to chat about Health+AI! 🧵👇

Chirag Nagpal

@nagpalchirag

2 years

1/n 📣 New Preprint: "𝑪𝒐𝒖𝒏𝒕𝒆𝒓𝒇𝒂𝒄𝒕𝒖𝒂𝒍 𝑷𝒉𝒆𝒏𝒐𝒕𝒚𝒑𝒊𝒏𝒈 𝒘𝒊𝒕𝒉 𝑪𝒆𝒏𝒔𝒐𝒓𝒆𝒅 𝑻𝒊𝒎𝒆-𝒕𝒐-𝑬𝒗𝒆𝒏𝒕𝒔" w/ @MononitoGoswami @AutonLab and @DrDufendach @UPMC #MachineLearning #EpiTwitter #MedTwitter #causalinference #DataScience

2

15

32

1

3

12

Chirag Nagpal

@nagpalchirag

4 years

NO "Data Scientists" and "AI Researchers", fitting exponential models or creating fancy #COVID19 dashboards is NOT, repeat NOT going to help the pandemic. Stop the Damn Hysteria! What might:

1

12

Chirag Nagpal

@nagpalchirag

4 years

@QueerinAI @AnimaAnandkumar @iclr_conf Totally love how the time is mentioned in IST 😍 . Seeing this for the first time in AI/ML Venues . Welcome how @QueerInAI is setting new standards for progressiveness in our community 😊.

0

2

11

Chirag Nagpal

@nagpalchirag

2 years

@MaartenvSmeden Wanted an office in @SCSatCMU ’s fancy gates hillman building

2

0

11

Chirag Nagpal

@nagpalchirag

4 years

🪔🪔🎇🎆Happy Diwali Everyone ! 🎆🎇🪔🪔

0

11

Chirag Nagpal

@nagpalchirag

5 years

Great to have ⁦⁦ @MFiterau ⁩ from ⁦⁦ @umasscs ⁩ back on home turf at ⁦⁦ @SCSatCMU ⁩, talking about her recent work in #ML4H #MLHC

0

1

11

Chirag Nagpal

@nagpalchirag

4 years

Many congratulations to @devikamalik_394 !! 👏🏽 for being nominated as @ForbesAsia #30under30 who’s doing amazing work in #disabilityinclusion through the #WheelingHappiness foundation ... setting lofty standards for our family!

Devika Malik 💜

@devikamalik_394

4 years

Made it on @ForbesAsia #30under30 list 2020, for the #disabilityinclusion work we have been doing at #WheelingHappiness foundation for half a decade!✨Wonderful recognition & silver lining in these times! Meet the #ForbesU30Asia 2020 class: #ForbesUnder30

5

40

2

3

10

Chirag Nagpal

@nagpalchirag

4 years

Hi @NarangVipin how much time of your life have you physically spent in South Asia? Would be good to know considering you claim to be South Asia strategy expert.

3

1

10

Chirag Nagpal

@nagpalchirag

6 months

It's a big day here at @GoogleAI @GoogleDeepMind with the Launch of Gemini ! Huge congrats to all friends and colleagues who've been involved and successfully delivered on this massive effort. :-)

0

10

Chirag Nagpal

@nagpalchirag

3 years

How has no one noticed that @Meta logo is exactly Microsoft @maluubainc logo!!!

0

10

Chirag Nagpal

@nagpalchirag

3 years

On this day, in a miraculous experiment, the Indian people gave themselves their constitution to become the world’s largest democracy! Happy Republic Day!!! 🇮🇳

1

0

10

Chirag Nagpal

@nagpalchirag

3 years

📢📢AAAI-Spring Symposium on Survival Analysis 📢📢 📢📢PhD Student Discussions - Call for Interest📢📢 #MLHC #ML4H

2

1

10

Chirag Nagpal

@nagpalchirag

5 years

Happy to be part of the groundbreaking TRACIR () project, joint effort between @SCSatCMU and @UPMCnews . As part of TRACIR, I would continue exploring more applied aspects of my research in ML for Rapid Minimally Invasive Hemorrhage Detection. #ML4HC

Pitt, CMU to Create an Autonomous Robotic Trauma Care System

Pitt and CMU scientists are teaming up to create an autonomous trauma care system that will fit in a backpack.

www.upmc.com

0

9

Chirag Nagpal

@nagpalchirag

4 years

Had a productive weekend. My conda environment finally resolved.

2

0

9

Chirag Nagpal

@nagpalchirag

4 years

Paper: This is joint work with Dennis Wei, @bhanuk31 , M. Shekhar, S. E. Berger, S. Das, and @krvarshney during my @IBMResearch @MITIBMLab 'Science for Social Good Fellowship'. It was a great experience collaborating with such a dynamic team!

Interpretable subgroup discovery in treatment effect estimation with application to opioid prescr...

dl.acm.org

1

3

9

Chirag Nagpal

@nagpalchirag

3 years

4) Permitting student organizations including, the Indian Graduate Student Association @cmuigsa amongst others to contribute parts of their annual budgets to support healthcare efforts.

1

4

8

Chirag Nagpal

@nagpalchirag

5 months

What if CS PhD programs required students to publish at least 1 solo author paper as a pre-requisite to graduate?

1

0

9

Chirag Nagpal

@nagpalchirag

5 years

getting a paper accepted in a theory conference is priceless, for everything else, theres a gpu.

1

0

9

Chirag Nagpal

@nagpalchirag

1 year

🐪🐪 Outside my grandparents house in Mhow, Madhya Pradesh, 🇮🇳 @MPTourism

1

0

9

Chirag Nagpal

@nagpalchirag

2 years

Delighted to see our Deep Survival prediction methodologies being applied in the real world. In this Scientific Reports paper, Dai et. al demonstrate effectiveness of DSM in estimating mortality risk of patients with localized prostate cancer.

Survival analysis of localized prostate cancer with deep learning

Scientific Reports - Survival analysis of localized prostate cancer with deep learning

www.nature.com

1

9

Chirag Nagpal

@nagpalchirag

4 years

If I could go back in time, I’d tell my previous self to choose venues I submit my work to more carefully. Good work can be completely missed if not made visible to the right audience. #AcademicChatter #AcademicTwitter

1

0

9

Chirag Nagpal

@nagpalchirag

5 years

High time Latex PDFs have Animated GIF support

2

0

8

Chirag Nagpal

@nagpalchirag

1 year

@Lan_Dao_ building a 30 storied building with only a few inches margin of error and ensuring it stands through tornados, earthquakes seems pretty darn hard to me. my vote goes to civil engineers.

3

0

9

Chirag Nagpal

@nagpalchirag

3 years

Congrats, scientific community! Deep Learning is now a religion!

Ilya Sutskever

@ilyasut

3 years

if you believe in deep learning, deep learning will believe in you!

47

141

1K

0

9

Chirag Nagpal

@nagpalchirag

3 years

The Biden administration’s complete lack of support to India in this desperate time of need is deeply disappointing.

0

2

9

Chirag Nagpal

@nagpalchirag

5 years

Unpopular Opinion: Indians are undrreprsented. The Indians you come acrss in STEM progrms in the US aren't reprsntative of the mch largr distribtn. They are sm supr talnted stdents, who would nevr get a chnce to be in a hlf decent US PhD progrm, only bcoz they didn't go to an IIT

0

9

Chirag Nagpal

@nagpalchirag

5 years

@gneubig Well that explains the nine lives 😅🤔

0

9

Chirag Nagpal

@nagpalchirag

4 years

Thank you so much for calling out the hypocrisy with which the west eulogies the devil called Churchill Prof @AnimaAnandkumar . You have the support of all of us!

0

8

Chirag Nagpal

@nagpalchirag

3 years

Last week I went to look for ghosts in the haunted ruins of #Bhangarh . I only found a lot of monkeys. @my_rajasthan

3

1

8

Chirag Nagpal

@nagpalchirag

4 years

Hard to believe that this innocuous picture, that was taken less than 12 months ago, is less than a couple of miles away from an almost active warzone. Hope the two sides resolve differences amicably and go back to maintaining peace.🇮🇳 #IndiaChinaTension #indiachinaborderconflict

1

0

8

Chirag Nagpal

@nagpalchirag

5 years

My research, long with other @IBMResearch Science for Social Good projects was featured on the @datanami blog!

BigDataWire

@BigDATAwireNews

5 years

Exploring new treatments for cancer. Tracking infectious disease emergence. Identifying pathways out of poverty. These are some of the #bigdata projects in @IBMResearch 's Science for Good campagin

0

4

1

0

2

8

Chirag Nagpal

@nagpalchirag

4 months

@nanjiang_cs weighted sum over all channels

0

8

Chirag Nagpal

@nagpalchirag

4 years

TIL about Poondi Kumaraswamy, a Civil Engineer who proposed what is now known as the Kumaraswamy distribution. This interesting distribution has support over [0,1] and a closed form PDF, CDF and Median.

1

8

Chirag Nagpal

@nagpalchirag

3 months

@latentjasper ... FrAcTaL KeRnElS ...

0

8

Chirag Nagpal

@nagpalchirag

5 years

Earlier this summer I visited #Ladakh #Leh @JandKTourism . Here is a picture of me riding a Bactrian Camel in #Nubra . I can confirm equitation skills transfer well, even in the presence of significant covariate shift. (Horse -> Camel)

1

0

7

Chirag Nagpal

@nagpalchirag

4 years

Having fun at @LTIatCMU 's Virtual Graduation with @shikibmehri talking about this years amazing Masters Cohort ! #LTIGrads #CMUgrads

0

1

7

Chirag Nagpal

@nagpalchirag

5 years

If ever your depressed about your research not going anywhere, remember, It's 2019 and we've still not figured out what does the fox say? #phdchat

1

7

Chirag Nagpal

@nagpalchirag

3 years

I love programming in python and its mostly perfect except: 1) class methods and `self` design choice is pretty bad. 2) `super` method is `super` confusing especially if you are switching between python versions. I hope python 4 fixes these two major poor design decisions.

1

0

8

Chirag Nagpal

@nagpalchirag

4 years

Wondering why training LSTMs with variable length sequences is still so hard in 2020..... AI is stupid if it can’t even learn how to pad and batch itself. 😤

1

0

8

Chirag Nagpal

@nagpalchirag

5 months

Some pictures from death valley national Park

0

7

Chirag Nagpal

@nagpalchirag

4 years

There is a very small difference between being a cofounder and a confounder.

0

7

Chirag Nagpal

@nagpalchirag

4 years

Using CNNs on Chest X-Rays to diagnose pneumonia and other adverse lung conditions is great and useful, but please don’t call these models “COVID detectors”. That’s deliberate misinformation that fuels public hysteria and is irresponsible academic practice.

0

8

Chirag Nagpal

@nagpalchirag

4 years

As of writing this tweet, the US has >1000 confirmed cases while India has <100. This #COVID19 visualization sucks. Why have a visualization at all, if it is not remotely accurate? Reduce the damn bandwidth of the smoothing kernel.

0

1

7

Chirag Nagpal

@nagpalchirag

3 years

1/ I think one needs to appreciate the difference between 'Algorithmic Fairness' from 'Ethical Machine Learning'. The former involves microeconomic constructs and techniques to measure disparities while the latter is broader involving anthropology, sociology and political science

1

3

8

Chirag Nagpal

@nagpalchirag

3 years

1/ So I interviewed with this hedge fund once and the interviewer asked me the formula for the conditional multivariate normal during the interview. I told him I don’t memorize things I can read from chapter 13 of Jordan’s book. Ofcourse, I didn’t get the internship.

2

0

7

Chirag Nagpal

@nagpalchirag

5 years

ok, so I jst realizd my name is in the latest @scikit_learn distribtion source. . Apprently an old ca. ~2016 PR from my undrgrad was mergd recntly . I have gotten a masters and a year into my PhD since then, yet feels super cool 😎

Add example of inductive clustering · Issue #4587 · scikit-learn/scikit-learn

Clusters can be found within a training dataset (transductive) then used to train a classifier, so as to extend the clustering model to new instances. See snippet at #4543 (comment) @selwyth is app...

github.com

2

0

7

Chirag Nagpal

@nagpalchirag

2 years

3/n Our proposed Approach "𝘾𝙤𝙭 𝙈𝙞𝙭𝙩𝙪𝙧𝙚𝙨 𝙬𝙞𝙩𝙝 𝙃𝙚𝙩𝙚𝙧𝙤𝙜𝙚𝙣𝙤𝙪𝙨 𝙀𝙛𝙛𝙚𝙘𝙩𝙨" involves learning representations of the input confounders followed by latent assignments to Z and Φ. We propose a tractable EM approach to perform inference under such a model.

1

2

7

Chirag Nagpal

@nagpalchirag

4 years

I’m sold.

Andreas Zeller

@AndreasZeller

4 years

What's it like to be a #PhD student in #Germany ? You can get paid, and well. You can afford a car, an apartment, and provide for a family. You may work with great advisors at great institutions. And the food... well, the food. Read on!

60

476

2K

1

0

7

Chirag Nagpal

@nagpalchirag

4 years

I have a joke about Non Negative Matrix Factorization, but forgot the Principal Component.

0

7

Chirag Nagpal

@nagpalchirag

3 years

What’s with clubhouse and gpt3 ? All of a sudden their seems to be 100s of AI startup ceos whose business has been revolutionized by gpt3! They all claim to have direct access to folks at open AI and musk? What’s going on?

3

1

7

Chirag Nagpal

@nagpalchirag

4 years

Happy Independence Day 🇮🇳 !

0

7