Prof. Dr. Marius Kloft

Marius Kloft is a professor of machine learning and deep learning at TU Kaiserslautern. He is an expert on unsupervised deep learning (particularly anomaly detection and deep generative models), multi-label classification, transfer learning, adversarial learning, explainable AI, and integration of heterogeneous data types in machine learning. He received the Google Most Influential Papers Award, the DFG Emmy-Noether Career Award, and the ANDEA Test-of-time Award for the most influential paper in anomaly detection in the last ten years (2012-2022). His award-winning paper 'Deep One-class Classification' was the first paper published at a top-tier machine-learning venue that uses deep learning for anomaly detection, a field now known as deep anomaly detection. Marius Kloft is spokesperson of the DFG Research Group FOR 5359 'Deep learning on sparse chemical process data'.

Contact

Prof. Dr. Marius Kloft University of Kaiserslautern-Landau Department of Computer Science Machine Learning Group Building 36, Room 325 Paul-Ehrlich-Straße 36 67663 Kaiserslautern 0631 205 2635, 0631 205 3286 ml@cs.uni-kl.de
Prof. Dr. Marius Kloft kloft@cs.uni-kl.de 0631 205 2800 Publications

Marius Kloft and group are interested in theory and algorithms of statistical machine learning (especially deep learning) and its applications. Their research covers a broad range of topics and applications, unifying theoretically proven approaches (e.g., based on learning theory) with recent advances (e.g., in deep learning or reinforcement learning). Topics they have been working on include unsupervised deep learning (particularly, anomaly detection), multi-modal learning, extreme classification, adversarial learning, explainable AI, and applications of ML in the life sciences, mechanical and chemical process engineering, and text analysis. Kloft has been serving as associate editor for journals (JMLR, TNNLS) and area chairs of conferences (AAAI, AISTATS, and ECML). Marius Kloft is spokesperson of the DFG FOR 5359 and co-organizer of the SPPs 2331 and 2364.

Marius Kloft has been working on applications of machine learning and deep learning in computer vision, natural language processing, computer security, plant research and agriculture, biomedical applications, mechanical engineering, and chemical process engineering. He co-developed the REMIND intrusion detection system and the CLEF 2012 Photo Annotation Challenge-winning image categorization system. As a former postdoc at Sloan-Kettering Cancer Center in New York, he is experienced in applications in the biomedical domain and oncology (e.g., analysis of histopathological, CT, and NMR data using deep learning and GWAS). Since joining TU Kaiserslautern in 2017, he has become increasingly engaged in mechanical and chemical engineering applications. For instance, he is the spokesperson of the DFG research group FOR 5359 ("Deep learning on sparse chemical process data"), the Carl-Zeiss project "Process Engineering 4.0", and he is a co-organizer of the DFG SPPs 2331 and 2364 at the interface of machine learning and chemical engineering. He is currently the primary PhD advisor of students at BASF and Bosch AI.

- Google Most Influential Papers Award (2013)
- DFG Emmy-Noether Career Award (2014)
- ANDEA Test-of-time Award (2022)

Basic Research
- Machine Learning (ML): Representation Learning, Zero-Shot/One-Shot/Few-Shot Learning, Self-Supervised Learning, Adversarial Learning, Reinforcement Learning (RL), Unsupervised Learning, Anomaly Detection, Density Estimation, Feature Engineering/Feature Extraction
- Robotics: Sensory Acquisition and Perception
- Technology Analysis: Economic Effects