4 ways to have fun with Machine Learning and Data
DKatalis ML Engineers share unique and entertaining ML projects you should try!
When we talk about ML/AI, the first impression that comes to mind would be complicated and serious projects. Let’s say creating a model to predict stock prices, diagnosing health issues, or detecting frauds.
But, there are always two sides to everything. Perhaps you don’t always have to utilize ML models for work, but also for entertainment and personal interests.
With the help of codes and AI, you can discover new ways to enjoy the activities that you like, and improve your skills at the same time! Because what better ways to learn than hands-on practice while having fun? You could also do some tweaks and then show off the projects as your portfolio later.
Here are some projects you should try, as recommended by DK’s ML Engineers:
Create a T-Shirt design using Deep Dream (Benjamin Tan Wei Hao)
DeepDream is a computer vision program that uses a convolutional neural network to find and enhance patterns in images via algorithmic pareidolia, thus creating a dream-like appearance reminiscent of a psychedelic experience in the deliberately over-processed images.
I wanted to come up with a T-shirt design that was:
- Colorful
- Unique
- Relatively low effort but still looked awesome.
There are online services that let you generate Deep Dream images, but most of them are not free, or only have low-resolution outputs. But with a little know-how, and a relatively capable GPU, it’s possible to do all these on your own!
So with the help of the designer, I took the base image and ran it through DeepDream. I iterated over a bunch of different hyperparameters and tried out 5 different Deep Learning models and generated around 3k images before settling on the final design.
What you need:
- This DeepDream repo by Aleksa Gordić
- An NVidia GPU (I used a GTX 1060 but a better one would give you high-resolution images)
- A decent, high-resolution input image to begin with.
Curious about our final T-shirt design using this method? Check it out 👇🏻
Pretty cool, right? Don’t forget to show us yours!
Analyze Your Relationship With ML (Varun Mallya)
This is a really fun project where Guy Tsror, one of the contributors on Medium, analyzed text messages with his girlfriend for a period of 1 year. The insights were quite fun, and to be honest, is a very nice geeky anniversary present.
But you could also try it with your best friends, siblings, parents, or even office friends!
This could be done by simply using text analysis tools which are very well documented in this medium article.
What you need:
- This code by Guy Tsror
- Your message history from messaging apps (the original project used WhatsApp)
- Python, pandas (or any data wrangling tools of your preference), data visualization tools.
There are many such examples of how this dataset can be processed: find out when is the prime time for you two to communicate? What are the most used words? How is the sentiment in general?
Check out more use examples here and here.
Create an AI friend
Do you want to have a chat with an AI “friend”? As interesting (or creepy) it might sound, this project is definitely cool and worth taking a look at.
It’s called CakeChat, which is basically a chatbot which you can have conversations with. The selling point? Your AI friend can show a range of emotions in their response to your prompts, be it anger, sadness, joy, or even fear.
You can get started by running their Docker images. There are separate versions for CPU and GPU.
For starters, you can try having simple conversations by playing with the APIs. And if you are feeling extra creative, go ahead and build a custom UI for this backend.
What you need:
Unique literary home decor using text analysis tools
In 2016, Adam Calhoun’s Medium post caught the public’s attention thanks to its unique analysis of famous authors’ works. Calhoun build a text analysis tool that
What can you do with the data? Plenty, of course!
Design unique room decors from your favorite texts with machine learning or just simply get a better understanding of your (or your favorite authors’) writing style!
It’s quite easy to do:
What you need:
- Python, this GitHub package by Adam J Calhoun.
- Text corpus from your favorite novels, stories, poems, or even your own writings!
Have other interesting ML projects to share? Let us know in the comments! In the meantime, follow us for more great recommendations from our engineers.
