I hope you had a great start to 2021. Welcome back to your monthly pulse on what I've been up to. I've given a few talks on QC-related things in the past month, and I built two projects in quantum classification. I also can't wait to share with you a gold podcast series I stumbled upon (the story of SpaceX vs Blue Origin); I know you're going to love it.
Richard Feynman once said, "If you think you understand quantum mechanics, then you don't understand quantum mechanics." What does this mean, and what happens when we do try to understand quantum mechanics? In this talk, I explore various interpretations of quantum mechanics and illustrate the power of asking good questions.
Inspired by a qiskit tutorial on quantum support vector machines, I went and built my own QSVM that classifies the malignancy of breast cancer tumours. Here, I explain my process on how I went about implementing this algorithm on the breast cancer dataset.
I took a curious dip into CRISPR applications in drug discovery and came across Synthego (genome engineering company). I thought that their solution to improve gene functional classification was cool so I made this video condensing what I learned.
I often find myself fantasizing about the future. What are the technological revolutions going to be? How will people's day-to-day lives change? In this episode, I shared my thoughts on what a day in the life of 2050 will look like, and the three main technologies that revolutionized life—gene editing, brain-computer interfaces, and QC.
Classification as a soft intro to quantum machine learning
How do you tell if a tumour is cancerous? How does your phone recognize your face? How does Gmail identify spam in your inbox? These are all classification problems that are suited for machine learning. In the past month, I got started with quantum machine learning (QML) by applying it to typical classification problems.
I'll describe my projects in further detail below but the bite-size version is that I built two models using qiskit (quantum computing framework for programming). One predicts breast cancer malignancy and the other determines wine quality 😄. Classification was interesting, but I wanted to move on to a more exciting area of QML—combinatorial optimization!
So nearing the end of the month, I started learning a famous QML algorithm used for combinatorial optimization—QAOA.
For fellow QC nerds that want to see some of what I've indulged in:
A support vector machine is a machine learning algorithm used for classification—in this case determining if a tumour is benign or malignant. A quantum support vector machine (QSVM) is similar to that but uses a quantum computer to find the perfect separation between the two classes (oversimplified). I implemented a QSVM for binary classification on the breast cancer dataset and then tested out different configurations in this project to observe deviations in accuracy.
After my project on binary classification, I implemented a multi-class QSVM on the wine dataset. And I ended up achieving an accuracy of 83% for the algorithm. The model is supposed to predict the quality of wine given certain attributes on it. This was a nice project to finish off my learning with quantum classification.
Wondery does a fantastic job of documenting the journey of each of these companies all the way from 2002 to now. This podcast series was so good I couldn't stop listening, I had no idea that each company had such an interesting backstory!
This video provides a solid, succinct summary of what antifragility is, and how you can apply it to your own life. I highly recommend checking it out, I'd even go as far to say that this is the most important determinant of success.
In the next month, I'll be attending Xanadu's hackathon on quantum machine learning so definitely looking forward to that! I'll also be going deeper into combinatorial optimization (QAOA most likely) and building more projects on it! Here's what you can expect for next month
At least two projects, one with QAOA on max-cut
Experimenting with Pennylane
Also looking forward to chip away at the QC portfolio optimization literature!
Quote I'm pondering...
"Learn the rules like a pro, so you can break them like an artist."
Thanks for tuning in,
P.S. There's so many different apple flavours, but only one apple juice flavour. Mind-blown? I was after first reading this. Take it as a bonus for peering down here ;) ✍❓