> [[xkcd]] 1838 So there's some task we want to solve. Instead of solving it manually, let's use a [[optimization|machine learning method]] and feed it [[dataset|data]] to solve the task automatically. ^goal # taxonomy So [[machine learning]] is all about automatically learning from data. Learning *what*? Generally speaking: - [[supervised]] learning makes use of *labelled* datasets to learn *prediction rules* $f : \mathcal{X} \mapsto \mathcal{Y}$. - [[unsupervised]] learning makes use of *unlabelled* data and tries to discover *hidden structure* (eg [[statistical inference]]). - [[reinforcement learning]] makes use of an *interactive environment* to learn *strategies* for accomplishing tasks. some other subproblems: 1. [[supervised]] - [[semi-supervised]] 2. [[unsupervised]] (and [[represent]]ation learning) - [[statistical inference]] - [[self supervised]] - [[autoregressive]] - [[contrastive learning]] 3. [[reinforcement learning]] - [[model based]] vs [[model free]] - [[on policy]] vs [[off policy]] - [[online reinforcement learning]] vs [[offline reinforcement learning]] - [[policy learning]] and [[approximate policy evaluation]] (or [[policy iteration|actor critic]]) other axes of classification: - non-probabilistic / probabilistic ([[Bayesian]]) - [[continuous optimization]] / [[discrete optimization]] https://paperswithcode.com/sota see also [[model assessment]] # sources this is a very miscellaneous list, more specific resources found on the relevant pages, also see [[deep learning]], [[research]], [[explore]], [[blogs and personal sites]], [[information retrieval and open resources and search engines]] - [Awesome Search](https://awesomelists.top/#/owainlewis/awesome-artificial-intelligence) (Mega list of AI resources, maybe go through it) ## courses - [Teaching - Dr. Raschka](https://pages.stat.wisc.edu/~sraschka/teaching/) (Courses at UW Madison) - [LinkedIn post about ML courses on YouTube, quite comprehensive](https://www.linkedin.com/posts/damienbenveniste_machinelearning-deeplearning-activity-6991448778692071424-y8op) - [crossposted from GitHub](https://github.com/dair-ai/ML-YouTube-Courses) - [Linear Algebra and Learning from Data](https://math.mit.edu/~gs/learningfromdata/) - [Courses – Stanford Artificial Intelligence Laboratory](https://ai.stanford.edu/courses/) - [CS 330 Deep Multi-Task and Meta Learning](https://cs330.stanford.edu/fall2021/) - [Lectures on YouTube](https://www.youtube.com/playlist?list=PLoROMvodv4rMIJ-TvblAIkw28Wxi27B36) - Seems neat! - [Practical Deep Learning for Coders - Practical Deep Learning](https://course.fast.ai/) - [[2022HardtRechtPatternsPredictionsActions|Patterns, Predictions, and Actions]] - [Fetching Title#qgcn](https://www.cs.ubc.ca/~nando/340-2012/lectures.php) - https://ml-visualized.com ## papers - [Weekly Papers](https://ai.papers.bar/papers/weekly) (AI Papers Bar) - [Machine Learning at Berkeley Blog](https://ml.berkeley.edu/blog/) - [Top Trending Computer Science & AI Papers](https://42papers.com/) - [Alpha Signal | The Best of Machine Learning. Summarized by AI.](https://alphasignal.ai/) - Twitter - [The latest in Machine Learning | Papers With Code](https://paperswithcode.com/) - EleutherAI Discord - [ōtoro.net](https://otoro.net/ml/) - hardmaru's ML research blog ## misc - [Tweet - "Here are some of my favorite OSS AI projects from this past week"](https://twitter.com/transitive_bs/status/1628065267514486789) (2023-02-21) - https://twitter.com/rasbt/status/1618973446066827264?s=20 - [Google's latest experiments in Labs](https://labs.withgoogle.com/) - [Courses – Stanford Artificial Intelligence Laboratory](https://ai.stanford.edu/courses/) - https://notesonai.com/ - [Toronto Intelligent Systems Lab](https://tisl.cs.toronto.edu/)