let me actually focus on one particular section of that text
Yes, a paper that was withdrawn from NIPS 2016 is one of these things that delays a book on Machine Learning for Artists. Artists are generally not the intended crowds of ArXiv preprints.
Here a second instance of the future staring at you in the eyes. It came as a tweet from Nick in the British understatement of the year this week
@OmicsOmicsBlog @BioMickWatson @OWallerman @lexnederbragt @Scalene Does this 771kb mappable read help at all? https://t.co/mtLbdHyWD8 pic.twitter.com/xE9DVJCMy7— Nick Loman (@pathogenomenick) 2 mars 2017
which led Clive Brown to state
Sequencing used to be about breaking things in small little pieces (100's of bases) and computations were required to put those information back together. Nick just called the reading of a piece with 700 000 bases! Sometimes, sensors can remove the need for computations in a major way. Check the previous discussions on the subject.
- Sunday Morning Insight: Escaping Feynman's NP-Hard "Map of a Cat": Genomic Sequencing Edition
- Sunday Morning Insight: What Happens When You Cross into P Territory ?
The third item can be found this paragraph from Google Research blog on a paper titled “Detecting Cancer Metastases on Gigapixel Pathology Images”.
...we showed that it was possible to train a model that either matched or exceeded the performance of a pathologist who had unlimited time to examine the slides....In fact, the prediction heatmaps produced by the algorithm had improved so much that the localization score (FROC) for the algorithm reached 89%, which significantly exceeded the score of 73% for a pathologist with no time constraint2. We were not the only ones to see promising results, as other groups were getting scores as high as 81% with the same dataset. Even more exciting for us was that our model generalized very well, even to images that were acquired from a different hospital using different scanners...
It's one thing to do better than humans, it's another to have a generalization across several different sensors. That last sentence is a confirmation that the technique isn't just an overfit of sorts.
Fourth item, Yaniv, who's been featured here on Nuit Blanche, (the last time was on Biological screens from linear codes: theory and tools) and Dina Zielinski have been working on using DNA for memory storage purposes.
Read our paper on DNA Fountain. Storing data on DNA. Shannon-capacity approaching, unlimited copies, 215PByte/gr. https://t.co/DAQ95Z26Hf pic.twitter.com/gtF1KdN7qk— Yaniv (((Erlich))) (@erlichya) 2 mars 2017
- an unheard of speed for ideas to be picked up by others
- a superhuman discriminative ability in disease classification using well known sensors
- an unheard of ability to decode the genome with a new sensor and a new way of encoding memories in that same genome.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.