Proud to announce our new paper, "Relevant Entity Selection: Knowledge Graph Bootstrapping via Zero-Shot Analogical Pruning" with Lucas Jarnac and Miguel Couceiro, accepted in #cikm2023

#knowledgeGraph #machineLearning #graphEmbedding #transferLearning #zeroShot #openScience

Contrarily to the hype, #AI is contributing to increase #carbonfootprint. I wrote a short summary, and proposed few ideas (sparse #neuralnetworks, simpler #LLM models, #neuromorphic hardware, #federatedlearning, #transferlearning)
https://alecrimi.substack.com/p/ai-carbon-footprint
#ml #sustainability

5 emerging trends in deep learning and artificial intelligence - Explore five emerging trends in deep learning and artificial inte... - https://cointelegraph.com/news/5-emerging-trends-in-deep-learning-and-artificial-intelligence #reinforcementlearning #federatedlearning #transferlearning #explainableai #deeplearning #transparency #dataprivacy

Data-Efficient #AI: How to Build Good AI Solutions When Data Is Scarce - https://t2m.io/XKpr3bFQ - #DeepLearning #TransferLearning #ContrastiveLearning

Learning fast and slow 🤖🧠⏳

Highly excited that our (long-term) work on more flexible #ReinforcementLearning and #TransferLearning is finally public.

Two separate processes enable fast, coarse adaptation and slow, but better final performance.

https://lnkd.in/dU9fFz4t

Learning fast and slow 🤖🧠⏳

Highly excited that our (long-term) work on more flexible #ReinforcementLearning and #TransferLearning is finally public.

Two separate processes enable fast, coarse adaptation and slow, but better final performance.

https://lnkd.in/dU9fFz4t

Trying to get a #TransferLearning #CNN to play nice with an already augmented image dataset. Consistently getting that horrid training/validation loss gap though. Think I'm going to have to run my own augmentation pipeline on the baseline images instead.

Yuk.

#MachineLearning #NeuralNetworks

#Newpaper: Our work for #nlproc on segmenting conversational data is now out on @NeuripsConf

We find that reusing pre-trained feature hierarchy in a #transferlearning paradigm proves to be ineffective for topic segmentation tasks on unstructured chat data.

Topic Segmentation in the Wild:Towards Segmentation of Semi-structured & Unstructured Chats w. Harjeet S, Sharanya K, Dhuri S, Samyadeep B, Soundar S, #Umass #Microsoft
📝Paper: https://arxiv.org/abs/2211.14954

Please DM to meetup @ #NeurIPS2022 :)

Our new work on segmenting conversational data is out on @NeuripsConf
Topic Segmentation in the Wild:Towards Segmentation of Semi-structured & Unstructured Chats https://arxiv.org/abs/2211.14954
w. Harjeet S., Sharanya K., Dhuri S., Samyadeep B., Soundar S., #UMass
#Microsoft #NeurIPS2022
We find that reusing pre-trained feature hierarchy in a #transferlearning paradigm proves to be ineffective for topic segmentation tasks on conversational data.

Please DM to chat meetup and chat more
#nlp

Want to share a milestone on a #ml #project I’ve been working on for a while on the side. I’m working on a #computervision application for #chess to detect the game state from a photo.
I used #transferlearning to fine tune a #convolutionalneuralnetwork with a new head for #regression to predict the four corners of the board.
I found synthetic datasets online (~6k images) and labeled ~1k real photos. I trained the model with a mix of random augmentations and projections with small errors.

Detect possible interactions between drugs using out-of-the-box Relation Extraction #SparkNLP model.

Live-demo: http://hubs.li/Q01sLGy_0

#datascience #textmining #healthcareai #deeplearning #machinelearning #artificialintelligence  #transferlearning #healthcaredatamining #nlp #nlproc

Cross-protein transfer learning substantially improves zero-shot prediction of disease variant effects

https://doi.org/10.1101/2022.11.15.516532

#disease #variantInterpretation #mutation #transferLearning

Continuing to share some of this year’s efforts: the publicly available free *Open World Lifelong Learning* course.

There are 12 #lectures , 18+ hours video, 600+ slides, to get anyone who is a bit familiar with #MachineLearning get started on #ContinualLearning #LifelongLearning .

It’s a broad introduction from #TransferLearning to continual forgetting, #ActiveLearning & #CurriculumLearning to #OODDetection , evaluation & more.

Web: https://www.owll-lab.com/teaching/cl_lecture/

Videos: https://www.youtube.com/playlist?list=PLm6QXeaB-XkA5-lVBB-h7XeYzFzgSh6sk

Today's workload:

Pull and review papers on the #ethics issues arising from #data driven business models for a 1000 words on implementing an ethics audit of info management strategy.

See if I can get a #CNN model with #TransferLearning included to differentiate images of monkeypox from chicken pox to a basic functioning level.

Looking forward to getting stuck into both thankfully!

For the gory details: we have used #machinelearning, specifically convolutional neural networks (CNN) pretrained on generic image recognition tasks and then adapted using #transferlearning to the problem domain of visual programming language identification. [2/n]