@XiaoMaterialLab

Vision Scientist and Associate Professor in CS at @AmericanU.

 Washington DC, USA
sites.google.com/site/beixiao/
Joined April 2023
  • Tweets97
  • Following151
  • Followers124
11Media
How is vision and language related in human material perception? New paper from my lab is now out at @PLOSCompBiol. We find verbal descriptions capture material appearance at a coarse level, but precise alignment is still lacking. @liao_oial @mswym8 journals.plos.org/ploscompbi…
3
9
20
1,731
2/2 Using an unsupervised alignment method (Gromov-Wasserstein Optimal Transport), we show the similarity structures between visual judgment and verbal description align on the coarse categorical level but lack precise one-on-one stimulus-level mapping.
1
106
1/2. Comparing representations of vision and language we find colorfulness, material name, and softness are the key features across participants. We also found that removing the “material names” from the text embeddings significantly decreased the correlation.
61
🚨 Medical AI Research alert! 🚨 Can LLMs predict clinical trial outcomes from protocol documents? American University Presents, CTP-LLM: Clinical Trial Phase Transition Prediction Using Large Language Models by @SaulehSiddiqui @XiaoMaterialLab A Large Language Model & Benchmark for Clinical Trials Here's why it's exciting: 1/9 👇🧵 #ClinicalTrials #MedicalAI #NLP #AI #Bioinformatics #GenAI #MachineLearning #HealthTech #Medical #LLM #GPT #Healthcare #pubmed #arxiv #vision #nlp #research #pathology #Radiology
2
11
1
33
2,193
Analyzing the image representation of materials obtained from various pre-trained deep neural networks, we find that similarity structures in human visual judgments align more closely with those of the vision-language models than purely vision-based models.
1
129
Very happy to share that our paper "CTP-LLM: Clinical Trial Phase Transition Prediction Using Large Language Models" is accepted by IEEE BIBM . arxiv.org/pdf/2408.10995 We predict #ClinicalTrials phase transition outcome from protocols using fine-tuned LLM.
1
1
5
602
Our first model, CTP-LLM, is a specialized version of GPT-3.5 Turbo, trained on our PhaseTransition (PT) Dataset. Our second model, BERT+RF, combines a clinical Bidirectional Encoder Representations from Transformers (BERT) with a Random Forest (RF).
1
78
I’m over the moon happy to celebrate the promotion to tenure of my colleague & office neighbor Hendrikje Nienborg @nienborglab. Her promotion shows that rigorous, careful science is rewarded, which is especially welcome news in today’s world of flashy but evanescent sound bites
Michael F. Chiang, MD@NEIDirector26 Aug 2024
Congratulations to Hendrikje Nienborg (@nienborglab) for receiving tenure at @NatEyeInstitute (@IRPatNIH @NIH)! In 2019, Hendrikje established a group that studies visual decision-making. More about her work here: go.nih.gov/6LbirYx
1
6
32
2,770
🚨 Paper submission alert for BMVC 2024! 📅 Mark your calendars — the final deadline is this Friday, 10 May 2024. Don't miss your chance to showcase your latest research in computer vision.🔍✨ #BMVC2024 #ComputerVision #PaperDeadline
1
8
24
4,719
Capital Graphics workshop on new research on #ComputerGraphics at @GeorgeMasonU Arlington Campus today. cragl.cs.gmu.edu/capitalgrap…
2
148
Bei Xiao retweeted
umdnacs@umdnacs8 Feb 2024
🧠At NACS Seminar tomorrow, Dr. Hendrikje Nienborg (@nienborglab) will tell us about "Perceptual decisions as action-perception loops in neural circuits." The talk is at 10:15am in BRB 1103. Please check the flyer below for more info!
1
1
4
1,249
Trying all day to upload a paper to biorxiv. The server doesn't work. Does anyone else also have this issue?
1
154
Finally succeeded around 10pm.
69
Someone wrote us to encourage the nomination of a female scientist for YIA for @VSSMtg . To be impactful within 10 years of PHD. One needs 1) a supportive (optimal) family, 2) effective mentoring, and 3) supportive institutes. All posed extra challenges for female scientists.
1
3
150
Of course, the investigator has to be excellent. But it does need a ton of luck and support. How do we give this support in earlier days? From day one?
77
I am very happy to share that #americanuniversity are among the first to receive an NSF ART grant Congratulations @dianaburley, @JosephKYoung, @DrRubidium, @SusannaCampbell, @DrRubidium @ButternutSquash, and Jordan Tama. Excited to be part of this team. new.nsf.gov/news/nsf-announc…

NSF announces first-ever Accelerating Research Translation awards to empower academic institutions...

The U.S. National Science Foundation today announced the first-ever Accelerating Research Translation (ART) investment — more than $100 million to 18 teams at academic institutions across the nation…

nsf.gov
6
626
Congratulations to team @dianaburley, @JosephKYoung, @DrRubidium, @SusannaCampbell, @XiaoMaterialLab, @ButternutSquash, and Jordan Tama for their receipt of this NSF ART grant! #americanuniversity #researchacceleration
U.S. National Science Foundation
@NSF15 Dec 2023
.@NSF's first-ever Accelerating Research Translation awards were just announced – an investment of more than $100M at academic institutions across the country to accelerate and scale translational research and grow the economy. bit.ly/3GI8Gpy
A graphic of a circle and interconnecting lines over a rough map of the U.S. and the text "ART"

ALT A graphic of a circle and interconnecting lines over a rough map of the U.S. and the text "ART"

5
13
2,245
This work is so cool! diffusionillusions.com/
1
125
This paper is so cool: dangeng.github.io/visual_ana… It shows several kinds of illusions that I had never seen before (e.g., color inversion illusion). It's exciting to see more and more cases like this, where AI opens up new kinds of art, rather than only imitating old forms.
10
132
8
534
64,725
Thanks everyone for being part of #BMVC2023, main track has finished but we have an exciting workshop day tomorrow! Kudos to @PandJLive for the amazing event organisation!
1
8
29
4,082
Load more