@NHulkund Profile picture

Neha Hulkund

@NHulkund

PhD Student at @MIT_CSAIL

Joined September 2020
Similar User
Pratik Joshi photo

@Roprajo

Sharut Gupta photo

@sharut_gupta

Matthew McDermott photo

@MattBMcDermott

Sara Beery photo

@sarameghanbeery

Tom Hartvigsen photo

@tom_hartvigsen

Priya L. Donti photo

@priyald17

Alexis Ross photo

@alexisjross

Eleni Triantafillou photo

@Eleni30fillou

Shobhita Sundaram photo

@shobsund

Maggie Makar photo

@Maggiemakar

Kayo Yin photo

@kayo_yin

Hanjie Chen photo

@hanjie_chen

Valerie Chen photo

@valeriechen_

Sanja Fidler photo

@FidlerSanja

Georgia Gkioxari photo

@georgiagkioxari

Pinned

Excited to announce that I will be starting a CS PhD at @MIT_CSAIL in the fall w/ NSF GRFP funding, advised by @sarameghanbeery! After graduating this week with my BS+MEng from MIT, I'm going to be at @ETH_en this summer, please let me know any recommendations in the area!

Tweet Image 1

Neha Hulkund Reposted

Google vs Apple. Who has the better looking app icons?

Tweet Image 1

Neha Hulkund Reposted

What happens when models see the world as humans do? In our #NeurIPS2024 paper we show that aligning to human perceptual preferences can *improve* general-purpose representations! 📝: arxiv.org/abs/2410.10817 🌐: percep-align.github.io 💻: github.com/ssundaram21/dr… (1/n)

Tweet Image 1

Neha Hulkund Reposted

The BeeryLab is busy at @eccvconf today! @juliachae_ and @EdwardVendrow co-organized @CV4E_ECCV workshop, happening 8:30-1, where I'll moderate the panel and Kai van Brunt and @__justinkay will present fish counting in sonar and Jae Joong Lee will present 3D trees ...


Neha Hulkund Reposted

Synthetic data has huge potential to drive new improvements in training and evaluation for computer vision. Interested in learning more about advancements and challenges? Join us at the SynData4CV Workshop at #CVPR2024 tomorrow (June 18)! syndata4cv.github.io

Calling all #CVPR2024 attendees! Join us at the SynData4CV Workshop at @CVPR (Jun 18 full day at Summit 423-425) to learn more about recent advancements in synthetic data! Explore more: syndata4cv.github.io



Neha Hulkund Reposted

@CVPR AI for Conservation happy hour on Monday! Open to anyone working on or interested in the intersection of CV/ML and ecology, conservation, sustainability, climate, etc Hosted w/ @sarameghanbeery @timmhaucke

Tweet Image 1

Neha Hulkund Reposted

Delighted to share one of my favorite pieces of work, now published: PURPLE, a method to estimate disparities in the prevalence of underreported outcomes, in women’s health and healthcare more broadly. nature.com/articles/s4429…


Neha Hulkund Reposted

Had a fun time attending NYC Computer Vision Day + sharing my work on Amodal Completion via Progressive Mixed Context Diffusion #CVPR2024 Project page: k8xu.github.io/amodal/ Event: cs.nyu.edu/~fouhey/NYCVis… Thank you to the organizers and sponsors! 🗽

Tweet Image 1

Neha Hulkund Reposted

Struggling with object detection out-of-domain? We are excited to introduce Align and Distill (ALDI), a state-of-the-art method and unified benchmarking & implementation framework for domain adaptive object detection. 1/ 🌐: aldi-daod.github.io 📜: arxiv.org/abs/2403.12029

Tweet Image 1

Neha Hulkund Reposted

I'm hiring a postdoc at @UVA to edit and detoxify LLMs! This is an exciting opportunity to join a vibrant research community and to collaborate with @StevenLJohnson and @MaartenSap Feel free to get in touch and please help spread the word! tinyurl.com/yeywuu3b


Neha Hulkund Reposted

How do bacteria navigate environmental variation in human hosts? In our newest preprint, we find bacteria can use “permanent” solutions for temporary problems — mutation-mediated phenotypic switching! See: biorxiv.org/content/10.110…


Neha Hulkund Reposted

Prof. @Grant_Van_Horn from @UMassAmherst giving a guest lecture in our @MITEECS Deep Learning course on developing machine learning systems that scale to millions of users, like the amazing work he has done with birdsong ID in @MerlinBirdID with @CornellBirds 🐦‍⬛🦢🦆🦜🦅🪿🦩

Tweet Image 1

Neha Hulkund Reposted

Presenting UniverSeg: an in-context learning model for medical image segmentation to appear at #ICCV2023 🎉! (w/ @jjgort, @mertrory, @AdrianDalca, and others) @MIT_CSAIL, @MIT, @MGHMartinos 🧵1/N (project-page, demo, and paper links 🔗at the end)

Tweet Image 1

Neha Hulkund Reposted

Personalized models often use group attributes like sex/age/race. In our latest w @berkustun @MarzyehGhassemi, we show how personalization can lead to worsenalization by reducing performance for some groups PDF bit.ly/43yump #ICML23 Oral bit.ly/3Kck7rQ

Tweet Image 1
Tweet Image 2

Neha Hulkund Reposted

I'm recruiting two PhD students to join my group @UVA to work on ML/AI/Health 🎉 Please share and come chat at #ICML2023! tomhartvigsen.com/join


Neha Hulkund Reposted

How will ML models perform when encountering new environments? A new work accepted to @TmlrOrg from @learn_ng, @NHulkund, @kchonyc & #JameelClinic PI @MarzyehGhassemi aims to predict how well ML models can generalize to ensure they can be safely deployed: openreview.net/pdf?id=jYkWdJz…

Tweet Image 1

Neha Hulkund Reposted

Large pre-trained models misbehave in deployment but are often too expensive to retrain, so we need ways to make targeted edits Presenting GRACE, a new method for making thousands of sequential edits that retains pre-training knowledge and prior edits! arxiv.org/abs/2211.11031

Tweet Image 1

Neha Hulkund Reposted

Excited to share that our work on predicting OOD generalization has been accepted to TMLR! (now ft. a new name, and updated experiments and analysis!) Joint work with @NHulkund, @kchonyc and @MarzyehGhassemi

Predicting Out-of-Domain Generalization with Neighborhood Invariance Nathan Hoyen Ng, Neha Hulkund, Kyunghyun Cho, Marzyeh Ghassemi. Action editor: Vincent Dumoulin. openreview.net/forum?id=jYkWd… #classifier #classification #generalization



Neha Hulkund Reposted

What makes two images look similar? Excited to share NIGHTS, a synthetic image similarity dataset with human judgements, and DreamSim, a human-aligned perceptual similarity metric. All you need is a pip install! Check it out here - dreamsim-nights.github.io

We’re releasing a new image similarity metric and dataset! --> DreamSim: a metric which outperforms LPIPS, CLIP, and DINO on similarity and retrieval tasks --> NIGHTS: a dataset of synthetic images with human similarity ratings paper+code+data: dreamsim-nights.github.io 1/n

Tweet Image 1


Neha Hulkund Reposted

T-1 hours until our Workshop on Robustness in Sequence Modeling starts in Room 290! We've got a stacked lineup of speakers, spotlights, and posters today, and it all starts with a great morning keynote by @bneyshabur!

Tweet Image 1

Loading...

Something went wrong.


Something went wrong.