Dr. @alpanrav
, our Chief Scientist – AI/ML, presented two of our AI solutions, Cough for TB and Loss to follow-up (LFU) and mortality prediction at the 36th @StopTB Partnership Board Meeting in Varanasi. @MoHFW_INDIA @TbDivision #WorldTBDay #YesWeCanEndTB #G20India #EndTB
MACHINE LEARNING
-
AI Solutions for TB Detection and Mortality Prediction Presented
By
–
-
AI-Powered Military Drones Transform Battlefield Operations
By
–
In this 'Ask INDIAai' series, checkout some amazing responses on the future of military operations with AI-powered drones! From surveillance to combat missions, these cutting-edge drones are changing the game on the battlefield. Read more: https://
bit.ly/40kIr0H -
MIT AI Model Detects Melanoma from Smartphone Pictures
By
–
Early detection of skin cancer thanks to #AI
— Pascal Bornet (@pascal_bornet) 27 mars 2023
MIT researchers have trained this machine learning model using 20,400 patient's skin pictures. The system achieves high precision and sensitivity in detecting melanoma… just from smartphone pictures!#techforgood #deeplearning pic.twitter.com/tWROH5PzwJEarly detection of skin cancer thanks to #AI MIT researchers have trained this machine learning model using 20,400 patient's skin pictures. The system achieves high precision and sensitivity in detecting melanoma… just from smartphone pictures! #techforgood #deeplearning
-
Internet Text Data Loss: Missing Draft Revisions for GPT Training
By
–
Yep! The interesting part is that most of the text on the internet is the "final" text, after you've revised it for a bit. All of that "latent structure" of your drafts, edits, going back and forth etc. is sadly lost. This would make for ideal data for GPTs so they can learn the
-
Deploy H2O Wave Apps to Hugging Face Spaces in Minutes
By
–
With #HuggingFaceSpaces, you can create and deploy ML-powered demos in minutes. Check out the latest blog from @unusual_code showing how easy it is to deploy your first @h2o_wave app into Spaces! #MachineLearning #ML #AI #DataScience @h2oai
-
AI Models Need Human Expert Validation Always
By
–
Perhaps, there's some errors, it's not because there is a reasoning, that it's always true, and the GPT, OPT, or other will succeed 100%, it's not the case at the moment. Human expert will always have the job to check for me. In French we have say : "l'erreur est humaine".
-
Knowledge Experience Creativity: Building AI Through Practical Application
By
–
Knowledge Vs Experience ⠀
> Knowledge is about collecting and connecting existing ideas⠀
> Experience stresses practice, or the application of knowledge over a prolonged period of time⠀
> Thanks to experience, comes creativity… #inspiration -
Research Community Blurring Facts and Fiction Lines with NLP
By
–
This is what I worry about. The lines between facts and fiction are being increasingly blurred, but if these lines are being blurred by the research community themselves, we have a huge problem! #NLP #GPT4 @Microsoft
-
AI hype fatigue drives interest toward alternative technologies
By
–
I wish we could turn down the AI/ML hype just a little bit. Maybe I should focus on something calmer, like blockchain.
-
Debugging ML Model Training: Shape Mismatch Resolution Success
By
–
About 4 hours and 30 minutes of intense debugging. Finally managed to make it work. Still gotta make considerable architectural changes, and make sure it *converges* But hey! at least it is training and not throwing shape mismatches anymore.