The 25 most important mathematical definitions in data science, v/
@DailyDoseOfDS_
.
@mit_csail
-
25 Essential Mathematical Definitions for Data Science
By
–
-
Knuth on Code Verification: Proving Correctness vs Testing
By
–
"Beware of bugs in the above code; I have only proved it correct, not tried it." — Donald Knuth, Turing winner & author of "The Art of Computer Programming"
-
Empty Object Cookies: A Programming Technique Explained
By
–
May I interest you in some empty object cookies? v/
@jeffistyping -
LLM Hallucination Leaderboard: Which Models Perform Best
By
–
Which LLMs hallucinate the most? The least? A leaderboard for how often these models produce hallucinations when summarizing a document: https://
bit.ly/4gyXsDG -
MIT Professor Uses AI Predictions Enhance Road Safety Accidents
By
–
MIT professor Hari Balakrishnan on how he's used Al predictions to make roads safer from accidents: https://t.co/rcxsIvsFHo pic.twitter.com/FkEAQTMMhX
— MIT CSAIL (@MIT_CSAIL) 19 décembre 2024MIT professor Hari Balakrishnan on how he's used Al predictions to make roads safer from accidents: https://
bit.ly/3P6tfjy -
iNaturalist Develops AI Query System for Scientific Image Discovery
By
–
The researchers are working w/iNaturalist to develop a query system to better help scientists & other curious minds find the images they actually want to see. Their working demo allows users to filter searches by species, enabling quicker discovery of relevant results like, say,
-
INQUIRE Dataset: Domain-Specific Image Search Challenges
By
–
What makes the INQUIRE dataset so challenging? The answer: its domain specificity. INQUIRE includes search queries based on discussions w/ecologists, biologists, oceanographers, & other experts about the types of images they’d look for, including animals’ unique physical
-
VLMs Ranking Images: INQUIRE Dataset Tests SigLIP Performance
By
–
They first used the INQUIRE dataset to test if VLMs could narrow a pool of five million images to the top 100 most relevant results (AKA ranking). For straightforward search queries like "a reef w/manmade structures & debris," relatively large models like "SigLIP" found matching
-
Multimodal LLMs Struggle with Reranking Task Performance
By
–
But how well could multimodal models rerank those 100 results? In these tests, huge LLMs trained on more curated data like GPT-4o struggled. Its precision score was only 59.6%, the highest achieved by any model.
-
Larger VLMs Risk Producing Biased Results for Nature Researchers
By
–
In these evaluations, the researchers found that larger, more advanced VLMs, which are trained on far more data, can sometimes get ecologists & other nature researchers the results they want to see. pic.twitter.com/B0uIYYj8M3
— MIT CSAIL (@MIT_CSAIL) 17 décembre 2024In these evaluations, the researchers found that larger, more advanced VLMs, which are trained on far more data, can sometimes get ecologists & other nature researchers the results they want to see.