As a class, we are working towards March 7th, where everyone will share a project that applies concepts from the first half of the semester (pre-trained models with ml5.js, TensorFlow.js, transformers.js). While training your own model is not required for this project, this week's assignment asks you to imagine an application that would.
Watch Machine Learning for Human Creative Practice by Dr. Rebecca Fiebrink (Eyeo 2018). Reflect on the question: How can machine learning support and expand creative practices?
Read A Brief History of Neural Nets and Deep Learning by Andrey Kurenkov.
Write a short blog post describing a creative application that requires training a custom model. Address the following:
What is the input?
What data will your model take in? Consider sources such as sensors, body/face/hand keypoints, images, sound, or text.
What is the output?
What will your model predict or generate? Are you classifying labels, predicting continuous values, or something else?
What kind of learning task is it?
Is this a classification or regression problem? It's okay if you're unsure!
What challenges do you anticipate?
These could be technical, ethical, or creative obstacles related to data collection, bias, implementation, or user interaction.
If you're feeling inspired, start prototyping your idea! You could create a p5.js sketch that collects sample data or explore how a user might interact with your system before training the model.
GPT Assist: