# From Quantum Theory to Published Research: My Journey into Quantum ML

Research Paper: "Quantum Kernel-based Support Vector Machine with Quantum Approximate Optimization Algorithm Embedding for Improved Lung Cancer Prediction"

Presented at: 2024 International Conference on Computer, Communication and Network Technologies (ICCCNT), IIT Mandi

This is the story of how a college student with no quantum background ended up presenting quantum machine learning research at a prestigious conference.

## The Beginning: Complete Confusion

I first heard about quantum computing in a random YouTube video. It sounded like science fiction:

Qubits that exist in multiple states simultaneously
Quantum entanglement
Exponential speedup over classical computers

I was hooked, but also completely lost.

## The Learning Curve

### Phase 1: The Basics (Month 1-2)

I started with the fundamentals:

Linear algebra: Vectors, matrices, eigenvalues (suddenly my math courses made sense)
Quantum mechanics basics: Superposition, entanglement, measurement
Quantum gates: X, Y, Z, Hadamard, CNOT

Resources that helped:

IBM Quantum Learning
Qiskit textbook
MIT OpenCourseWare on Quantum Mechanics
YouTube channels (3Blue1Brown for intuition)

It was hard. Really hard. I must have read about superposition 20 times before it clicked.

### Phase 2: Quantum Programming (Month 3-4)

Theory was one thing. Actually programming quantum circuits was another.

I learned Qiskit (IBM's quantum programming framework):

<code/>
from qiskit import QuantumCircuit, execute, Aer

# Create a quantum circuit with 2 qubits
qc = QuantumCircuit(2)

# Apply Hadamard gate to create superposition
qc.h(0)

# Create entanglement with CNOT
qc.cx(0, 1)

# Measure
qc.measure_all()

# Run on simulator
backend = Aer.get_backend('qasm_simulator')
job = execute(qc, backend, shots=1000)
result = job.result()
counts = result.get_counts()

Seeing quantum circuits actually run (even on simulators) was magical.

### Phase 3: Quantum Machine Learning (Month 5-6)

This is where it got interesting. Quantum computers could potentially speed up machine learning algorithms.

Key concepts:

Quantum kernels: Using quantum states to compute similarity between data points
Variational quantum algorithms: Hybrid quantum-classical approaches
QAOA (Quantum Approximate Optimization Algorithm): For optimization problems

## The Research Project

### Finding the Problem

My professor mentioned that traditional machine learning for medical diagnosis could be improved with quantum approaches. Lung cancer prediction seemed like a perfect use case:

Large datasets
Complex feature relationships
Critical need for accuracy

### The Idea: QSVM + QAOA

Our approach combined:

Quantum Support Vector Machines (QSVM): Use quantum kernels to find better decision boundaries
QAOA for feature embedding: Optimize how features are encoded into quantum states

The hypothesis: Quantum feature spaces could capture patterns that classical methods miss.

### Implementation Challenges

#### Challenge 1: Noisy Simulations

Real quantum computers are noisy. Even simulators model this noise.

Solution: We implemented error mitigation techniques and ran multiple shots to average out noise.

#### Challenge 2: Limited Qubits

Each feature needs qubits. We had to be strategic about feature selection and encoding.

Solution: We used feature engineering to reduce dimensionality, then quantum embedding to expand it in a quantum-friendly way.

#### Challenge 3: Classical vs Quantum Performance

Initial results showed quantum was slower and not more accurate!

Solution: We tuned hyperparameters, improved our quantum circuit design, and focused on specific data subsets where quantum advantage was clear.

### The Breakthrough

After weeks of tuning, we found it: For certain feature combinations and dataset sizes, our quantum approach achieved:

2-3% higher accuracy than classical SVM
Better generalization on unseen data
Interesting insights into feature relationships

Was it revolutionary? No. But it was proof that quantum ML could work for real problems.

## Presenting at IIT Mandi

Getting to present at ICCCNT 2024 was surreal.

Preparation:

Practiced the presentation 50+ times
Created visualizations to explain quantum concepts simply
Prepared for tough questions from quantum physics experts

The Presentation:

15 minutes to explain months of work
Audience of professors, researchers, and PhD students
My hands were shaking, but I made it through

The Questions: Some were easy ("Why QAOA specifically?"), others were brutal ("Have you considered topological qubits?").

But I survived. And I learned more from those 30 minutes of questions than from weeks of research.

## What I Actually Learned

### Technical Skills

Quantum circuit design
Qiskit programming
Classical ML algorithms (for comparison)
Statistical analysis
Academic writing

### Bigger Lessons

1. You Don't Need to Know Everything to Start I started with zero quantum knowledge. Six months later, I'm presenting research. Don't wait until you feel "ready."

2. Research is Mostly Failure For every experiment that worked, 10 didn't. That's normal. The trick is learning from each failure.

3. Collaboration Multiplies Impact I couldn't have done this alone. My co-authors brought expertise in medicine, statistics, and quantum theory.

4. Quantum Computing is Real It's not just theoretical anymore. Companies like IBM, Google, and IonQ have real quantum computers you can program right now.

5. The Field is Wide Open Quantum ML is so new that a college student can contribute meaningful research. There's room for everyone.

## The State of Quantum Computing (2025)

Reality check: We're not replacing classical computers tomorrow.

What quantum computers can do:

Solve specific optimization problems
Simulate quantum systems
Certain ML tasks (experimentally)

What they can't do (yet):

Run Crysis
Replace your laptop
Solve P vs NP

When will they be useful:

Niche applications: 1-3 years
Broader ML applications: 5-10 years
General purpose: 15-20 years

## Getting Started with Quantum Computing

Want to explore quantum? Here's my recommended path:

### 1. Prerequisites

Linear algebra (mandatory)
Python programming
Basic probability
Classical ML (helpful but not required)

### 2. Learning Resources

IBM Quantum Learning (free, excellent)
Qiskit Textbook (comprehensive)
Microsoft Quantum Katas (hands-on practice)
Pennylane (for quantum ML specifically)

### 3. First Projects

Build a quantum random number generator
Implement Grover's algorithm
Create a simple quantum classifier
Simulate quantum teleportation

### 4. Going Deeper

Take online courses (edX, Coursera)
Read research papers
Join quantum computing communities
Contribute to open-source quantum projects

## What's Next for Me

I'm continuing in quantum ML, exploring:

Quantum neural networks
Quantum generative models
Applications in drug discovery
Error correction techniques

And I'm looking for collaborators! If you're interested in quantum computing, let's talk.

## Final Thoughts

Quantum computing feels like the early days of classical computing. We're figuring out what it's good for, building the tools, and discovering the possibilities.

It's the perfect time to jump in. The field needs:

Programmers who can build quantum algorithms
ML engineers who can find quantum applications
Researchers who can push the boundaries
Communicators who can explain it to everyone else

You don't need a PhD. You don't need to be at MIT. You just need curiosity and persistence.

The quantum future is being built right now. Why not help build it?

Interested in quantum computing or want to discuss research ideas? Connect with me on LinkedIn or email me at mohanasusm1@gmail.com.

## Additional Resources

Our research paper: [Link to paper]
My quantum ML code: [GitHub repo]
Recommended reading list: [Blog post]
Quantum computing Q&A: [Twitter thread]