ML-PROTOTYPING
In this four-day workshop we will show you how to quickly prototype your ideas by implementing supercharged if-then scenarios with machine learning technologies. After shortly introducing some basic concepts of machine learning and how creatives can leverage the technology for their own process, we will dive into developing ideas and tinkering with the algorithms. Due to many registrations we may run short in supplies, so if you own any of these, please remember to bring them:
→Your laptop
→Raspberry Pi
→MicroUSB chargers
→MicroUSB and ethernet cables
→Servo or continuous motors
→Printer filament compatible with the university printers
→Similar tinkering supplies you like to work with
→Glue, scissors, cutterknives, some paper etc.
→IMPORTANT: You will need a blynk account and a new version of the chrome browser (>v75) installed. Please make an account and download the browser before the workshop starts.
We expect all participants to be present the entire time more or less. As the curriculum of the HfG makes it basically impossible to schedule workshops for more than four days, we will try to squeeze everything into this short time frame. This may result in some open end working. We want everybody to be able to produce presentable outcomes, so we will also insist you are motivated to present, share and document your projects. We will meet in the KVB at 10 am. Depending on the internet connection we will need to move to the PC-Pool in the Isenburger Schloss.
THURSDAY
10:00
〉 short introduction to machine learning basics.
〉 interesting ML projects on the web.
〉 some experimenting and getting to know the tools
〉 introduction to the workshop topic, groups
FRIDAY
10:00
〉 working on concepts and narrowing down by experimenting
13:30
〉 introduction to the IOT-Toolkit we will use
〉 set-Up of our IOT-Hubs
〉 brainstorming ideas and short presentation
13:30
〉 presentation of two ideas
〉 feedback and making plans
〉 experimenting
〉 shopping lists
SATURDAY
10:00
〉 shopping for missing parts
〉 working on the projects
SUNDAY
10:00
〉 project work
〉 finalization
15:00
〉 present your progress, hurdles, complications and solutions
〉 project work
13:00
〉 finalization
〉 pictures
〉 final presentation
SCRATCH YOUR ITCH
to scratch one's own itch:
»To do something out of motivation to solve a personal problem; to take matters into one's own hands.
〉Brief:
Come up with inventive ideas how to solve your individual problems with the help of machine learning algorithms. You can begin with brainstorming daily itches like cold coffee and forgetting your keys. If you want you can expand this thinking to more serious problems like work place safety, injury prevention or inclusive product ideas. How can these soft if-then cases help you, a person you know or a fictional person to scratch their itch?
You can try to come up with something really helpful, but if you want to go with a humorous (not silly!) or very individual approach, feel free to do so.
Example: Jeff suffers from cold feet when he spends long winter days in the office. With the help of an image recognition algorithm, he trained a model to detect when his feet are under the table to turn on a small infrared heat lamp.
Example two: After a injury Jill suffers from a short-term loss of hearing. To not miss the doorbell, she trains an algorithm to turn the lights red, when the door bell is heared.
〉Questions to consider after brainstorming and narrowing down the project:
〉 How can the product be presented? Is it possible to give it an imaginary shape? How would it look? For prototyping purposes we distribute the processing power with our laptops, but in a real world product, the TPU-Board would have a size of a RaspberryPi.
〉 How does the person interact with the product? How is it made visible to other people other than it's inventor? Can it be reprogrammed to serve additional use cases? Is it necessary to retrain it, for example if you cut your hair or grow a beard, have a cold (different voice)?
〉 If it doesn't have a shape: How can the functionality be showcased? Try to use as few different forms of media as possible. iE three to four pictures, two GIFs, one GIF + a mock-up.
Goal of our Workshop presentation is to show how machine learning algorithms could be used to create a new type of product. We will help you to find a coherent narrative in the different exhibits which highlight the different concepts you come up with. We will publish a microsite with the workshop outcomes for you to reference in your portfolio.
Find the classifiers at:
PoseNet pose classifier
SoundRec sound and voice recognition
ImageRec image recognition