Yinjun Wu

Assistant professor, school of computer science, Peking University

[Yinjun Wu @wu_yinjun]

At ICML 2025 @icmlconf ? Don’t miss Poster Session 6 West #W-105! My student Junlin will present our latest work on vector search. Come by to chat, share feedback, or discuss ideas! #MachineLearning #VectorSearch

blog: https://t.co/o56MbWXCGx
Paper: https://t.co/uvIQ4EHkCJ

[Adam Stein @adamlsteinl]

Excited to present our ICML 2024 paper: "Towards Compositionality in Concept Learning"!
🔗 Blog: https://t.co/V8qM0iNXIN
📄 Paper: https://t.co/3ukytg3ziO
💻 Code: https://t.co/PZJO3BRBWq

[Eric Wong @RICEric22]

Traditional concept vectors used to explain deep representations fail to compose when combined, i.e.
🐤(small) +🦢(white) =🦩(big & colorful)❌

We propose CCE: a method for extracting *composable* concepts, i.e.
🐤(small) +🦢(white) =🕊️(small & white)✅

https://t.co/9u1xJQmygq

[Ted Satterthwaite @sattertt]

CONGRATS to Dr. @chenying_zhao, who received her PhD in Biomedical Engineering yesterday!!!!!!!!  Her software -- like ModelArray (https://t.co/lhkLjZgYbQ) & BABS -- for highly-reproducible + large-scale image analysis have changed the way we work. CONGRATS DR. ZHAO!!!!!

[Zack Ives @zgives]

We're excited to welcome our database community friends to Philly -- in person! -- in a few short months! Registration will be opening shortly, linked to https://t.co/J3g4L2v3T7. #SIGMOD22 @SIGMOD

[Mayur Naik @AI4Code]

Excellent talk by Molham Aref. The relational model could become a "foundation model" for neuro-symbolic AI. This model underlies our recent work on the Scallop language for neural logic programming (NeurIPS'21) and the CodeTrek model for code understanding (ICLR'22).

[VLDB 2021 @vldb2021]

Susan Davidson received the 2021 VLDB Women in Database Research Award!! Be sure to attend Susan's talk "It's not just Cookies and Tea" (3/4)

[PVLDB @pvldb]

Vol:14 No:11 → CHEF: A Cheap and Fast Pipeline for Iteratively Cleaning Label Uncertainties https://t.co/PIVntP2IvF

[Edgar Dobriban @EdgarDobriban]

Better late than never! Time to discuss our #icml 2020 paper DeltaGrad for rapid retraining of #ml methods (w/ @wu_yinjun & Susan Davidson). Do you ever need to train the same model after adding/deleting a few datapoints? See https://t.co/LWGSYD0kex & read on 1/

[Yinjun Wu @wu_yinjun]

\ue022

[Yinjun Wu @wu_yinjun]

2222222 ;D

[Yinjun Wu @wu_yinjun]

#pretend

[Yinjun Wu @wu_yinjun]

@justinbieber

[Yinjun Wu @wu_yinjun]

#Shoes

[Yinjun Wu @wu_yinjun]

cool

[Yinjun Wu @wu_yinjun]

#cool