I Algorithms: Machine Learning Cards K12 Edition I Welcome to you print-at-home version of the I Love Algorithms card deck. Simply print at home (color is best!) and cut out the cards to begin playing with algorithms. This deck can be used as-is, on its own, or go to https://dschool.stanford.edu/ emerging-tech to pair with other tools we're building. Spread the love... Please use this deck, share it, and credit us. We would like to know how you use it. For more info visit dschool.stanford.edu. Credits I love Algorithms Deck Carissa Carter, Megan Stariha Gameboard and Dataset Cards Megan Stariha, Asha Lamanque Ariam Mogos I love Algorithms K12 Edition Carissa Carter, Megan Stariha, Asha Lamanque, Ariam Mogos. Illustrations Carissa Carter Design Direction Daniel Frumhoff *Please do not repurpose without attribution. Adapted from the “I Love Algorithms Game” by the Stanford d.school. Licensed under CC BY 4.0. Algorithms! K12 Edition I Algorithms! I love...what? An algorithm is a piece of computer code that can take information and tell us more about it. There are a lot of different types of algorithms used in machine learning. Machine learning is a way to look at a lot of data where the computer learns and gets better at learning over time. This stack of cards describes six types of machine learning algorithms. Since humans like different ways of learning, this stack of cards explains machine learning algorithms in three different ways I Algorithms! Can my love run deeper than these six algorithms? Of course. This deck includes the basic types of machine learning algorithms, but there are many ways to expand on this content in future versions. One concept we have not covered here is deep learning. Deep learning is a type of machine learning where the computer learns what to do without being explicitly programmed to do so. It is the gateway to things like speech and image recognition, analysis of text, and much more. Neat! I Algorithms! So...why do I need to know anything about algorithms? In order for the technologies of today and tomorrow (and all of the things that they power) to represent all of us, they need to be built by all of us. You don’t need to be the coder, but you need to know what the code can do. If you understand what machine learning algorithms can do, you can better envision the implications of your designs. You can influence conversations about data and bias. I Algorithms! We hope you love this deck as much as you love algorithms! This project was created by the Stanford d.school. The d.school helps people develop their creative abilities. It’s a place, a community, and a mindset. As designers and educators, we believe in providing radical access to the intersection of technology and design. Spread the love... Please use this deck, share it, and credit us. We would like to know how you use it. For more info visit dschool.stanford.edu. I Algorithms! K12 Edition Association Association Association is when there is a strong relationship between two or more things based on a theme or idea. Association If someone likes to create short videos are they 80% more likely to have TikTok? If someone plays Minecraft are they 70% more likely to also play with LEGO bricks? Clustering Clustering Clustering is when the system groups things or puts them together based upon their similarities. Are any of these things related somehow? Clustering I need to organize my favorite songs, what types of playlists should I make? You can cluster according to genres of music like dance, hip hop, pop, rock, electronic. Dimensionality Reduction Dimensionality reduction is where important information is highlighted when there is a lot of data, so that we can focus and maintain simplicity. Dimensionality Reduction Dimensionality Reduction Can you just tell me what’s important in my data? Reinforcement Learning Reinforcement Learning Reinforcement Learning Put your machine into a place and give it a goal or a game. It starts to interact and try to figure out what it should do to achieve the goal. It really wants to win! This is a great algorithm for programming robots. “I do not need you humans to teach me, I learn by trying things on my own.” How do I win this game? Classification Classification Classification This algorithm predicts what category something might be put in. You (the human) give it lots of data, and you either tell it what categories to pick from, or let it figure it out itself! How might this car drive itself? How do I learn which youtube video to play next? Are you asking yourself… Am I eating strawberry jam or strawberry jelly? Is this a picture of a seal or a sea lion? Does the x-ray show that the student has a broken finger? Is that a vegetable or a fruit? Regression Regression is the relationship-finder. For finding connections between different things. It’s useful for predicting (like the weather) or for things where historical events help suggest what might happen in the future. Give it the information and example answers. It compares its answers with the right ones to get better. Regression Regression Are you curious about…. “I wonder how much my pokemon cards will be worth in 30 years?!” I Algorithms: Dataset Cards K12 Edition Disney Plus Spotify This dataset lists the shows on Disney Plus with information regarding their ratings and genres. Contains 1,000,000 playlists, including playlist- and track-level metadata (artist, songname, and song length). Baked in bias: This dataset contains primarily English language movies and shows. Cereals This dataset lists the nutritional value (calories, percentage of daily values, etc.) of major cereal brands along with how they are served. Baked in bias: Almost all of the cereals are gluten-based, there are very few gluten-free brands. Biodiversity Dataset The National Park Service publishes a database of species identified in individual national parks and details about the size of the park and its climate. Baked in bias: 60% or more of the songs and playlists are pop music. Baked in bias: This dataset is dependent upon individual national parks that may prioritize data collection differently, causing inaccuracies. Wikipedia Corpus Dataset Sesame Street Characters All text on Wikipedia. It contains almost 1.9 billion words from more than 4 million articles. Dataset of over 1800 Sesame Street characters from the television series, including their personality traits and interests. Baked in bias: 84% of Wikipedia editors are male. A majority are 17-40. Census/ Population Data The census dataset includes the age, sex, and occupation for 23,000 US households. Baked in bias: If people don't respond to the census, they aren't counted, and census data is only recorded every 10 years. Baked in bias: This dataset overly represents Sesame Street characters who are grumpy. CelebA Dataset This dataset contains more than 200,000 images of celebrity faces. Baked in bias: This dataset overly represents celebrities who are white and male.
