A practical LaTeX writing guide is available to help PhD students quickly self-check common submission mistakes and reduce repetitive work. ✅ Prompt LLMs for auto-checking ✅ Fast, easy, and effective! 🚪 Portal: https://t.co/VuFDoGqf3B Welcome to read, share, and contribute! 🙏

Excited to share that our latest paper on training-free adaptation for CLIP-style VLMs has been accepted to #ICCV2025! We propose Token Condensation as Adaptation, which improves performance with up to 48.9% lower compute — no training, no tuning! Check it out 👇

⏰ Upcoming SIGIR-AP 2025 Deadlines! Abstract due: July 1 Paper due: July 8 Soliciting research, resource, reproducibility, perspective, and industry papers. Had a paper rejected at SIGIR'25? Resubmit as SIGIR-RR or regular submission! More info:

🔍 [3/3] Color-Oriented Redundancy Reduction in Dataset Distillation 📍 Join us at the poster: West Ballroom A-D #7309 🗓️ Thu, Dec 12 | 11 a.m. PST — 2 p.m. PST Official page: https://t.co/1lurWgpUMb

Also happy to present 2 more data-centric works on retrieval and dataset distillation! 🌟 🔍 [2/3] Exploiting Descriptive Completeness Prior for Cross-Modal Hashing with Incomplete Labels 📍East Exhibit Hall A-C #1400 🗓️ Wed, Dec 11 | 11 a.m. — 2 p.m. PST https://t.co/Ukm5s6nEXe

Recoding and paper can be found here:

[3/3] Our solution? DiPEx, a self-supervised prompting solution for VLMs like GroundingDiNO progressively learns to expand a set of distinct, non-overlapping hyperspherical prompts to enhance recall rates. It shows +20% AR gain over existing prompting and SAM on COCO and LVIS! 🔥

[2/3] Class-agnostic object detection can be a cornerstone or a bottleneck for many downstream OOD and open-vocabulary vision tasks. However, it faces challenges in maintaining high recall rates, especially with the complexity of object types and contexts.

Excited to share our @NeurIPSConf work on unsupervised prompt expansion and stop by to meet our first author! 🚀 🔍 [1/3] DiPEx: Dispersing Prompt Expansion for Class-Agnostic Object Detection (by Jason Lim) 📍 Poster: East Exhibit Hall A-C #1300 🗓️ Fri, Dec 13 | 11 a.m.-2 p.m

Young scientists regularly ask me for career advice. Academia or industry? Big company or startup? US or Europe?  Good scientists in AI disciplines are fortunate to have many choices. But choosing can be stressful. I always give the same advice. 1/10

The #ACMMM24 website is now live! ACM MM is happening in Melbourne, Australia from 28 October to 1 November 2024. If you would like to attend, present or exhibit at the world's premier conference in multimedia, submit an expression of interest today:

#InMemoryOfDrYifanLiu #GoneTooSoon

Yifan was not only a gifted researcher but also a wonderful friend. Her passion for knowledge was truly inspiring. She will be deeply missed. My heartfelt condolences go out to her family and loved ones. 💔 #InMemoryOfYIfan

[UQ Data Science Seminar] Wed, 17/05/2023 - 13:00

[UQ Data Science Seminar] Wed, 10/05/2023 - 13:00 - Building Experiment Tracking at Scale with Weights & Biases -

[UQ Data Science Seminar] Wed, 03/05/2023 - 10:00 - SEINE: SEgment-based Indexing for NEural Information Retrieval -

[UQ Data Science Seminar] Wed, 22/03/2023 - 13:00 - The Human Quotient for Better AI Systems -

[UQ Data Science Seminar] Wed, 15/03/2023 - 13:00 - Multi-Domain Few-Shot Image Classification -

I and @Yadan_Luo are looking for an Australia-based PhD student to work on domain adaptation, who will be spending 1 year at IIT Delhi and rest of his/her PhD at the University of Queensland. Topic details: https://t.co/zlwi8jLN3t Apply:

[UQ Data Science Seminar] Wed, 01/03/2023 - 13:00 - Seven Algorithms for the Same Task (Testing Uniformity) - https://t.co/GJaGWogiU1 Zoom Link: