Mark D. McDonnell
Papers
2
Total Citations
9
H-Index
2
About
Mark D. McDonnell is a researcher whose work bridges neural network architecture and applied machine learning for sustainability. His primary research areas include efficient neural network design, rapid training algorithms, and automated material classification for recycling systems. McDonnell's most notable contribution is his work on modular expansion of hidden layers in single-layer feedforward neural networks, a 2016 paper that presents an architecture and training algorithm designed for exceptionally fast training with minimal computational resources. This approach, which constructively expands output weights, addresses critical challenges in deploying neural networks on low-power systems. In applied research, McDonnell has tackled robotic sorting of recycled beverage containers, using style-transfer techniques to improve material classification accuracy for glass, plastic, metal, and liquid-packaging-board. This work has direct implications for improving recycling efficiency and reducing contamination in waste streams. While his citation counts (7 and 2 for his top papers) reflect a focused rather than widely-cited portfolio, McDonnell’s contributions are significant for their practical orientation—developing computationally efficient methods that can be deployed in real-world, resource-constrained environments. His research demonstrates a commitment to making machine learning both faster and more applicable to pressing environmental challenges.
Research Focus
Key Achievements
Top Papers
- 1
- 2