There are usually fewer tweets on weekends. Luckily, we have some older tweets retweeted by people I follow.
I’ve added a very basic search page powered by Algolia. The free plan should get us covered for at least six months.
Rocket Lander
Rocket Lander OpenAI Gym environment inspired by SpaceX's Falcon 9 vertical landing, using Box2D physics engine. They experiment using PID, MPC control benchmarks, and also tried DQN, DDPG ... https://t.co/egb9iyCtPS pic.twitter.com/v4ihiTCgwa
— hardmaru (@hardmaru) May 27, 2018
If you are interested in using neural networks for challenging rocket control problems, here's a paper from 2003 that uses Neuroevolution to train a neural network for controlling a difficult “finless” rocket.
— hardmaru (@hardmaru) May 28, 2018
Slides (34-38): https://t.co/rB0ZF5XoV7 pic.twitter.com/E8LD3og2SF
“Active Guidance for a Finless Rocket Using Neuroevolution” (2003), by Faustino J. Gomez and Risto Miikkulainen. They won a best paper award at GECCO2003. Gomez went on to start @nnaisense with Schmidhuber, and Miikkulainen started https://t.co/uxSJNlC5dm. https://t.co/NAOY1a1OPj pic.twitter.com/v9DP7SCGAL
— hardmaru (@hardmaru) May 28, 2018
LIME on Random Forest
I played around with LIME on an advanced regression problem (house-prices) https://t.co/RKrpddrx1l
— Verena Haunschmid (@ExpectAPatronum) January 24, 2018
| paper & python package by @marcotcr | r package {lime} by @thomasp85 #rstats #machinelearning
Altair Tutorial
Video from my #PyCon2018 Altair tutorial is up! Notebooks are available here: https://t.co/ODZv0nFbzVhttps://t.co/6d4gqky1zl
— Jake VanderPlas (@jakevdp) May 12, 2018
Floating Numbers
The nicest visualization I've ever seen of how floating point numbers work: https://t.co/SUF8go07TI
— John Myles White (@johnmyleswhite) May 27, 2018
AI and Compute
AI and Compute: Our analysis showing that the amount of compute used in the largest AI training runs has had a doubling period of 3.5 months since 2012 (net increase of 300,000x): https://t.co/YH4tZXirtU
— OpenAI (@OpenAI) May 16, 2018
Sitting at the forefront of computing, we like to imagine we have a good grip on rates of progress. Understanding that Moore's Law - hailed as a miracle enabler of the computing era - is a _snail_ compared to AI's compute trend is vital. Thanks for raising this @OpenAI. https://t.co/S3eR30HrPy
— Smerity (@Smerity) May 16, 2018
This doesn't state that brilliant research can't arise from efficient hardware use, it's simply saying that if brute compute can be used advantageously for training an agent (search, self play, ...) then it will be. Compute will become engrained within and across devices.
— Smerity (@Smerity) May 16, 2018
Being guarded against hype is important - but being cognizant of how quickly the landscape may change in our field of endeavor (and the industries and communities tied to it) is equally important.
— Smerity (@Smerity) May 16, 2018
Machine Learning Specialization Course
Google Launches a new #MachineLearning Specialization Course: https://t.co/QvAXGcAUK3 #BigData #DataScience #DeepLearning #AI #TensorFlow #Algorithms
— Kirk Borne (@KirkDBorne) May 28, 2018
Miscellaneous
And again, this is such an important story. A software failure rate of 20% may sound fine to some, but that's 7000 students forced to leave the UK due to a false assessment https://t.co/uMGjBOibEH
— Kate Crawford (@katecrawford) May 1, 2018
lots of exciting recent work in large-scale distributed training of neural nets: (very) large-batch SGD, KFAC, ES, population-based training / ENAS, (online) distillation, ... 🔥
— Andrej Karpathy (@karpathy) May 27, 2018
Not the first time I've made and shared this list (and not comprehensive), but figured worth sharing again for anyone interested in alternatives to boxplots. #dataviz pic.twitter.com/rOG6UFGoQc
— Cole Knaflic (@storywithdata) May 16, 2018
Why start a course with datasets on day one?
— David Robinson (@drob) May 17, 2018
“I’ve heard students say ‘I hate math’ and ‘I hate stats.’ I’ve never heard one say ‘I hate data.’
-@minebocek at #SDSS18
"Are you embarrassed by the first iteration of one of your machine learning projects, where you didn’t include obvious and important features?"
— Trey Causey (@treycausey) April 24, 2018
Data scientists, don't overlook qualitative methods! https://t.co/Ml9cLGbZwS
Statisticians are called weird for terms like Law of the Unconscious Statistician (LOTUS), but I just heard that actuaries use "Darth Vader Rule" for the fact that the expectation of a nonnegative r.v. is the integral of its survival function. pic.twitter.com/mlKhMH86BI
— Joe Blitzstein (@stat110) May 27, 2018