Speaker
Description
While quantum computing has introduced new
directions for computational modeling, there remains limited
empirical understanding of how quantum-inspired models behave
relative to classical approaches under varying data availability
conditions. Most existing studies evaluate models under fixed
dataset settings, limiting insight into how performance evolves
across realistic data regimes.
This work presents a systematic benchmark of quantum and
classical models under controlled dataset size variations. The
study evaluates Support Vector Machines (SVM), Random Forest
classifiers, Quantum Support Vector Classifiers (QSVC), and
Variational Quantum Classifiers (VQC) using structured feature
representations extracted from plant leaf images. Instead of
raw image inputs, engineered features capturing color, texture,
and edge information are used to ensure a consistent and fair
comparison across model families.
Experiments are conducted across multiple dataset sizes to
analyze how performance scales with increasing data availability.
The focus is not on identifying a single best-performing model,
but on characterizing differences in stability, generalization
behavior, and sensitivity to data constraints.
Results indicate that performance trends depend strongly on
dataset size, with quantum models showing competitive behavior
under limited data regimes, while classical models demonstrate
more stable scaling as dataset size increases.
This work contributes a reproducible benchmarking frame
work for analyzing quantum and classical model behavior under
realistic data constraints, offering insights into model selection
under low and moderate-data conditions in quantum computing
research.
| Topic | Quantum algorithms for machine learning applications |
|---|---|
| Additional Topic | Hybrid quantum-classical approaches for learning and optimization |