IMA Low Res | NYU Tisch School of the Arts + NYU Shanghai
Make Art with AI is an introductory course for students with no prior machine learning knowledge to explore and utilize different experimental machine learning methods, such as ChatGPT, image generation (Stable Diffusion), music generation (AIVA), pose estimation, body segmentation, pitch detection and image & sound classification with k-nearest neighbors algorithm (KNN). Working with a variety of pre-trained models, students will design and code artistic, practical or interactive projects for the web. Diverse hands-on techniques with p5.js, ml5.js and tensorflow.js will be introduced with a large number of example codes so that students can directly apply the methods to their own ideas, solutions and thesis projects.
Upon completion of this course, students will be able to:
- recognize, categorize and utilize various Machine Learning (ML) models in ml5.js and expand their knowledge and experience in ML;
- review, practice and produce the fundamentals of programming and integrate the concepts into creative or practical applications;
- develop and explain a classification process with the KNN algorithm for unique, unconventional and meaningful interaction with the process;
- visualize and simulate the various data from ML models and demonstrate the effective uses of the ML models.
- apply diverse computational techniques with ML models to their own ideas and practices;
- create artistic, innovative, practical and/or interactive artifacts on the web platform with ML models, and;
- produce, demonstrate and evaluate generative art and/or creative applications by utilizing a combination of concepts and techniques discussed over the course.
Learning and teaching Machine Learning can be a daunting task. This class seeks to reverse the conventional methods of teaching Machine Learning by applying a more friendly and approachable style, masking the complexity of the concepts and technologies. Students will review the fundamentals of programming first. Then they will use existing machine learning models (pre-trained) and apply them to their own ideas and outputs, similar to the way we utilize physical sensors with Arudino or devices such as Leap Motion and Kinect without a full understanding of its construction or blueprint.
The courses will progress with three hands-on workshop-style sessions and one presentation/sharing session. By the end of the course, on the presentation/sharing day, students will complete and present their projects. Diverse techniques and applications of ML models will be demonstrated during weekly synchronous sessions. The instructor will facilitate programming concepts and techniques in live-coding. Students will be encouraged to actively follow the code along with the instructor, ask questions, and engage in both creative and technical discussions.
Students will also be encouraged to reach out to the instructor outside of class, and ask questions, share ideas/feedback and discuss the topics and techniques in detail.