Instead of sharing the usual weekend read, here is one or our papers for a change: Rank Consistent Ordinal Regression for Neural Networks. A short and concise method for using deep neural networks for ordinal regression tasks in a rank-consistent manner: https://t.co/jczRuEH0va pic.twitter.com/NDMgrFzkjU
— Sebastian Raschka (@rasbt) November 6, 2020
Apply {Torch} to your Tables!
— RStudio (@rstudio) November 5, 2020
"torch for tabular data" from the RStudio AI Blog. @zkajdan
on how to use #torch for deep learning on tabular data, including a mix of categorical & numerical features.
Also: how not to die from poisonous mushrooms.#rstats #datascience #ml https://t.co/6UEY2SM2EC pic.twitter.com/pIjJtkMwfB
In https://t.co/HhhpKEKrCM, @sinafazelpour & I showed how most ML fairness research consist of ideal theorizing, suffering all the pitfalls that generally plague that approach. Here, Daniel Estrada goes further, addressing ***why*** AI ethics researcher embrace this approach. https://t.co/XDPXNCg87G
— Zachary Lipton (@zacharylipton) November 5, 2020
"Monitoring Machine Learning Models in Production"
— ML Review (@ml_review) November 5, 2020
A Comprehensive Guide https://t.co/YMK8ddPMA6 pic.twitter.com/Y0pqlBUnx1
The cat and mouse games with large language models are going to be fascinating to watch. A recent example (of many) https://t.co/u2lhjwuvLZ if offense is sufficiently advantaged/strong (which I think is likely) then maybe we can't have nice things
— Andrej Karpathy (@karpathy) November 5, 2020
https://t.co/meGdKh3aY4 pic.twitter.com/Q3D5EbulCl
— Chris Albon (@chrisalbon) November 4, 2020
You can have B and C without A, e.g.:
— Yann LeCun (@ylecun) November 4, 2020
[Bromley et al. NIPS 1993] https://t.co/2ObnCppnyq
[Chopra, Hadsell, LeCun CVPR 2005]
[Hadsell, Chopra, LeCun CVPR 2006]
[Taulor et al. CVPR 2011]
[Taigman et al. CVPR 2015]
(and several others between then and now)
A. NCA is a loss.
— Yann LeCun (@ylecun) November 4, 2020
B. Siamese net is an architecture.
C. Contrastive learning is a general training paradigm to give good scores to observed things and bad scores to unobserved things.
Those are independent concepts.
You can have:
- C without A or B
- B without A or C (BYOL)
Rise of the Self-Replicators: Early Visions of Machines, AI and Robots That Can Reproduce and Evolve
— hardmaru (@hardmaru) November 4, 2020
Amazing and accessible book by @drtimt @AnimAlanDorin about the history and modern development of self-replicating machines!
Free PDF (consider buying)→ https://t.co/49eT6QjUuS pic.twitter.com/5rqmXkcc5y
“If we define intelligence as whatever is needed to create artificial intelligence, the answer seems to be that intelligence is all of civilization.” @togelius https://t.co/qHG8jhVqP4 https://t.co/GYmd3qtThy
— hardmaru (@hardmaru) November 4, 2020
Check out our recent piece on "Preparing for the
— Nicolas Papernot (@NicolasPapernot) November 2, 2020
Age of Deepfakes and Disinformation" for @StanfordHAI 's policy brief series cc @danboneh @GrottoAndrew https://t.co/Jlt4Fm5Ao7 pic.twitter.com/GJrFQ6lidI
In the last transformers release, we teamed up with @raydistributed @anyscalecompute to provide a simple yet powerful integration for hyperparameter tuning.
— Hugging Face (@huggingface) November 2, 2020
To demonstrate it, @richliaw shows you how to fine tune BERT on MRPC leveraging multiple GPUs. ⤵️https://t.co/TFaVKcRlme
