A Biologically-Inspired Quantum Machine Learning Algorithm

Overview

The biological brain is the most efficient computer in existence, performing 10 18 operations per second using only 20 watts of power. State-of-the-art machine learning lags exponentially behind in energy efficiency, optimal coding, and robustness compared to biological systems of similar scale. This project constructed a biologically-inspired machine-learning vision model that mimics the behavior of specific biological neural circuits, to improve model performance and efficiency. This project hypothesizes that a quantum approach will more aptly mimic biological behaviors and demonstrate increased benefits due to intrinsic quantum properties such as the global quantum entanglement for feature extraction and increased parallelization. The proposed model is comprised of two units. The first unit models simple and complex cells for preprocessing, and the second unit extracts abstract, global features of the image using a quantum kernel convolution, whose weights are trained with an Adam optimizer and cross-entropy loss function. The proposed model was compared to classical CNN and biological models. Models were trained on MNIST and Fashion-MNIST datasets. To test the similarity of the models to biological networks, they were presented with novel image distortions, such as noise, contrast modulations, adversarial attacks, and shape -vs-texture bias. Due to the high adaptability and robustness of the human vis ual system, augmented images are easily recognizable by human observers, but historically difficult for machine learning models. When tested, all models achieved similar raw accuracy, but the proposed quantum model surpasses other models in robustness and energy efficiency. This project concludes that quantum properties hold promise for biologically-inspired machine learning. Year 3: A Novel Water Filtration System Using Nano-Particle Enhanced Moringa oleifera and Coconut Shell-Activated Carbon Satvika Nadella Allen High School, Allen, TX According to the U.S. E.P.A., wastewater treatment facilities process approximately 34 billion gallons of wastewater daily. Once filtered, wastewater is released into local water sources despite containing levels of nitrogen and phosphate. Greywater is a c ommon wastewater type produced from households that includes types of soapwater. Nanoparticles have been recently studied as a new source of treatment for wastewater and propose an efficient method of filtration to efficiently purify heavily contaminated wastewater. This study aims to refine the previously established ability of the two -stage Moringa Seed and Coconut Shell -Activated Carbon (CSAC) Filter. By incorporating MgO nanoparticles into the system, we aim to enhance the moringa seed’s natural antibiotic effect to accelerate the filtration process. The two-stage testing process began with testing various MgO concentrations (25%, 50%, 75%) in the moringa pod to determine the optimal purification of lead, copper powder, ammonium nitrate, and phosphate, in simulated water. Results demonstrated 50% MgO allowing for complete removal of contaminants, which led to its use in subsequent testing. Stage 2 of testing incorporated MgO within the Moringa Seed and CSAC filter. This enhanced system was tested using greywater, and results demonstrated 100% elimination of contaminants and no observable bacterial growth over five days. Daphnia magna were also observed in the filtered water, and the heartrates were stable at 276 bpm, determining the water to be non-toxic. This novel, Nanoparticle-enhanced Moringa oleifera and Coconut Shell-Activated Carbon Filter is proven to be a low-cost, effective greywater filtration system applicable for global use.

Competition history

• JSHS 2025 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.