Congratulations to the top ten finalists of our #DataScience and #MachineLearning #Hackathon for #AI4Good & AI for India! Your innovative solutions are shaping a brighter future. Keep up the amazing work! #H2OWorldIndia #AirPollution #Environment #Kaggle #AI
MACHINE LEARNING
-
H2O AI targets Kaggle Grandmaster status within 12 months
By
–
"Some of my leadership team who are here today, I've made them feel very nervous by saying that in the next 12 months, I want to have our own #KaggleGrandmasters @CommBank and we're very confident that with the partnership that we can make that happen together."
-

CommBank Runs 1000+ Models on 157B Data Points for Customer Experience
By
–
"We run over 1000 models in real time, across 157 billion data points to make personal and relevant experiences available to our customers across all of our different channels." – Andrew McMullan, Chief Data & Analytics Officer @CommBank #CustomerObsessed #H2OWorldIndia
-

Google AI Releases Improved MC4 Corpus and uMT5 Models
By
–
Sharing a piece of work I contributed to while at @GoogleAI
: * a new improved Mc4 corpus (29T char tokens and 107 languages) that gets language sampling right with UniMax sampling. * open source pretrained uMT5 models trained on 1T tokens. * Unimax sampling solves some -

Self-Organizing Maps Solve Traveling Salesman Problem Efficiently
By
–
Traveling Salesman Problem is a well-known challenge which consists in finding the shortest route possible that goes through all cities only once. With 15 cities, that's 44 billion possibilities! A new approach using Self-Organizing Maps now solves it https://
diego.codes/post/som-tsp/ -
SpectFormer Improves Vision Transformers via Frequency Attention
By
–
Microsoft & Bath U’s SpectFormer Significantly Improves Vision Transformers via Frequency and Attention https://
syncedreview.com/2023/04/18/mic
rosoft-bath-us-spectformer-significantly-improves-vision-transformers-via-frequency-and-attention/
… -

H2O World Comes to India for First Time
By
–
H2O World comes to India for the first time, coinciding with the rise of AI and Data Science talent in the country! Read more here: http://
ow.ly/ovJa50NMnPm #AI #DataScience #H2OWorldIndia -
Scaling Laws Enable Predictable Model Performance
By
–
I don't think so? "Scaling laws" states that perplexity is highly predictable. For example, GPT-4's loss on some evaluations can be predicted with models of less than 1,000x compute. (
https://
arxiv.org/pdf/2303.08774
.pdf
…)
Should be a clear difference between things that can be predicted with -
Frustration over someone claiming algorithms are fake
By
–
Ummmm he just said “algorithms are fake” do I throw a chair?
-
Postdoc Position: NLP and Speech Processing for Police Relations
By
–
We have a postdoc position available! Come join our computational social science project on using NLP and speech processing on body-worn camera data to help improve police-community relations!
