The same boring projects than everyone else is working on. How do you break off the mold and make a difference?
Thread: 7 machine learning projects that will teach you the technical skills you need to succeed out there.
Today, companies are dumping insane amounts of money on people with the right skills.
But here we are, showing up with the same "MNIST Digit Recognition" and "Iris dataset" experience.
These are good to start, but you need to take your learning to another level.
The good news for you: Most people can't be bothered.
Most people will bookmark this thread and will never do anything with it.
Even if you do one of these 7 projects, you'll be doing more than 99% of everyone who reads this!
It has never been easier to stand out!
Why should you care about solving these projects?
1. They will give you experience. 2. They will stretch your abilities. 3. They will force you to learn. 4. They will show off beautifully in your portfolio. 5. They will make you stand out.
The first one is critical!
What makes these projects "interesting"?
They aren't the same projects that everyone is working on.
These are hard but doable. They are different, and they match the skills that companies are hiring for.
These projects will get you a job!
Here are your 7 projects:
1. Detect movement in a video 2. Stock trading bot 3. Track a person throughout a video 4. Speech emotion recognition 5. License number plate recognition 6. Customer segmentation 7. Identify the language used in a text
Let's now talk specifics.
First, I'm not adding a lot of details on purpose.
These are open-ended ideas that you can take whereever you want, depending on your skills and interest.
Creativity is a feature.
Where can you get the data for these?
• 1, 3: You won't need any data.
• 4, 6: Link to datasets below.
• 2, 5, 7: Easy to find online.
You are going to need a few things to solve these problems. Some suggestions:
1. Python 2. Deep learning 3. Regular machine vision algorithms (OpenCV library) 4. Facebook's fastText model 5. Reinforcement learning
This is not an exhaustive list.
Something to keep in mind:
There's no "Solving these problems at 100%." These are hard, and there's no end to any of these.
But, you can definitely come up with a solution that's good enough for a controlled scenario. That's a win!
For example, building an automatic trading bot for one stock that beats one arbitrary index is possible.
Doing that for any stock is not that easy.
Detecting *every* language out there? Yeah, that's hard too.
A rough list of steps:
• Pick a problem
• Define your boundaries
• Break it down in smaller pieces
• What's out there that can help you?
• Learn what you need
• Build as much as you can
• Publish it online for others to see
Important: write about what you learned.
I post threads like this every week.
Stay tuned as I help you get to the core of practical machine learning.
You can find the rest of my threads here: @svpino.
TL;DR:
7 machine learning projects you can build:
1. Detect movement in a video 2. Stock trading bot 3. Track a person throughout a video 4. Speech emotion recognition 5. License number plate recognition 6. Customer segmentation 7. Identify the language used in a text
For automatic trading bots, there's a course in Udacity called "Machine Learning for Trading."
1. Recommendation system for e-commerce 2. House Price prediction 3. Music Genre detection 4. Sentiment Analysis from tweets 5. Titanic Survival ← Start here.
This is a great idea, for sure! You have “unlimited” high-quality images to play with.
1. You need a lot of math to start 2. You need a Ph.D. to get a job 3. You always need a lot of data 4. You need to buy expensive hardware 5. It's hard to become proficient in it 6. It's the solution for most problems
Bullshit.
In the last 6 months, I've posted more than 100 threads here on Twitter talking about machine learning and how you can build a career on it.
And I'm just getting started!
Stay tuned. A lot more is coming.
First misconception: All machine learning is hardware-hungry.
Deep learning stretches you, but outside that, it gets much better.
If you need GPUs/TPUs, there are many free/cheap options you can use, especially while learning.