It’s easy to associate AI and Machine Learning with technology, programming, and the future, but today’s applications range from medicine to e-sports. Keep in mind that AI and ML have taken over much of the finance and investment world over the past decade. Explore today’s innovations for yourself in the two links below (both from April, 2019):
- AI powerhouse IBM already uses AI to accurately predict when employees are considering quitting the company. A similar system has been developed to predict whether hospital patients will be readmitted within 30 days. The big commonality in these applications is that the prediction relates to INDIVIDUALS, not groups as previous statistical approaches could do. One big difference between the applications is that IBM collected data in real time from current employees and has acted already to keep the workers. The hospital application was developed and tested ONLY on data from clinical notes. Here’s the link: https://venturebeat.com/2019/04/11/ai-predicts-hospital-readmission-rates-from-clinical-notes/
- OpenAI is an advanced AI development group whose AI has just defeated a team of the best players of the e-sport game Dota-2. The AI learned to play the complex strategy game on its own. This quote from the developer captures the key message: “OpenAI Five is powered by deep reinforce learning, which means we didn’t code it how to play. We coded it how to learn.” Their next goal is to teach the AI how to play in collaboration with humans. Here’s the article: https://www.theverge.com/2019/4/13/18309459/openai-five-dota-2-finals-ai-bot-competition-og-e-sports-the-international-champion
One of most futuristic plans in human and machine intelligence comes from Neuralink, founded by Elon Musk to explore the development of a “neural lace.” Neural lace is a direct link of high-performance technology to the brain as imagined in Ian Banks’ science fiction work, Surface Detail. Researchers associated with Neuralink have released a paper describing a machine to implant brain electrodes — sort of a brain “sewing machine.” Here’s a brief description:https://futurism.com/the-byte/sewing-machine-implant-brain-electrodes
Dataset Sources to Build & Test Your Own Machine Learner
Kaggle (below) has both datasets and problems that go with them. Also note the list of datasets preinstalled with R. There are similar lists for scikitlearn and tensorflow
- R Datasets (preinstalled with normal installation of R). Each one includes description of variables and sample R code: https://stat.ethz.ch/R-manual/R-patched/library/datasets/html/00Index.html
- Google Dataset Search: https://toolbox.google.com/datasetsearch
- NOAA (the US National Oceanic and Atmospheric Administration) operates weather stations that measure surface temperatures at different sites around the United States. The hourly readings are publicly available: http://www.ncdc.noaa.gov/qclcd/QCLCD?prior=N
- U.S. Census Data:https://www2.census.gov/programs-surveys
- MovieLens / GroupLens movie recommender system. Research project with choice of recommender algorithms and links to open datasets (Google search listing): https://www.google.com/search?client=safari&rls=en&q=movielens&ie=UTF-8&oe=UTF-8
- Boston Housing Dataset:http://www.cs.toronto.edu/%7Edelve/data/boston/bostonDetail.html
- Kaggle.com –The Home of Data Science:http://kaggle.com maintains over 14,000 TAGGED datasets useful to bring into an R or Python environment to explore. A free account is required, then you can access the datasets at: https://www.kaggle.com/datasets
- 19 Free Data Sets (many are highly useful and current: https://www.springboard.com/blog/free-public-data-sets-data-science-project/
- Gapminder Data Sets (global datasets, not always current, but awesome. Especially for social sciences). https://gapminder.org/data