Krystian Radlak

Papers

Total Citations

H-Index

About

Krystian Radlak is a researcher at the forefront of deep learning safety and engineering education, whose work addresses critical challenges in artificial intelligence. His primary research areas include out-of-distribution (OOD) detection for deep neural networks (DNNs) and the integration of computer vision with robotics. Radlak’s most notable contribution, "Detection of Out-of-Distribution Samples Using Binary Neuron Activation Patterns" (2023), has garnered 14 citations, introducing a novel method to identify novel inputs that DNN classifiers fail to recognize—a crucial advancement for safety-critical applications. This work tackles a fundamental limitation of AI systems, enhancing their reliability in real-world scenarios. Additionally, his earlier paper, "Integration of robotic arm manipulator with computer vision in a project-based learning environment" (2015, 5 citations), showcases his commitment to innovative pedagogy. In this study, Radlak describes a Project-Based Learning (PBL) approach at Sogn og Fjordane University College in Norway, where undergraduate electrical engineering students built a robot capable of playing tic-tac-toe. This hands-on project not only taught technical skills but also fostered collaborative problem-solving. With a focus on both theoretical rigor and practical application, Radlak’s work bridges the gap between cutting-edge AI safety and accessible engineering education, making him a valuable contributor to the field.