Adaptive and explainable Video Anomaly Detection
Video Anomaly Detection (VAD) aims to automatically identify unusual events in video that deviate from normal patterns. Existing methods often rely on One-Class or Weakly Supervised learning: the former uses only normal data for training, while the latter leverages video-level labels. Recent advances in Vision-Language Models (VLMs) and Large Language Models (LLMs) have improved both the performance and explainability of VAD systems. Despite progress on public benchmarks, challenges remain. Most methods are limited to a single domain, leading to performance drops when applied to new datasets with different anomaly definitions. Additionally, they assume all training data is available upfront, which is unrealistic for real-world deployment where models must adapt to new data over time. Few approaches explore multimodal adaptation using natural language rules to define normal and abnormal events, offering a more intuitive and flexible way to update VAD systems without needing new video samples.
This PhD research aims to develop adaptable Video Anomaly Detection methods capable of handling new domains or anomaly types using few video examples and/or textual rules.
The main lines of research will be the following:
• Cross-Domain Adaptation in VAD: improving robustness against domain gaps through Few-Shot adaptation;
• Continual Learning in VAD: continually enriching the model to deal with new types of anomalies;
• Multimodal Few-Shot Learning: facilitating the model adaptation process through rules in natural language.
Physics-Informed Learning for Acoustic Inverse Problems: Field Reconstruction, Detection, and Detectability Analysis in Complex Environments
This PhD project aims to develop a mathematical and algorithmic framework for solving acoustic inverse problems in complex environments, based on physics-informed learning. By explicitly incorporating the wave equation into artificial intelligence architectures, the objective is to improve acoustic field reconstruction from partial measurements, the localization of mobile sources, and the quantitative analysis of their detectability. The project combines partial differential equation modeling, constrained optimization, and hybrid deep learning. Applications include distributed acoustic sensing systems and the detection of mobile platforms.
Development and validation of surface haptics machine learning algorithms for touch and dexterity assessment in neurodevelopmental disorders
The aim of this PhD thesis is to develop new clinical assessment methods using surface haptics technologies, developed at CEA List, and machine learning algorithms for testing and monitoring tactile-motor integration. In particular, the thesis will investigate and validate the development of a multimodal analytics pipeline that converts surface haptics signals and dexterity exercises inputs (i.e. tactile stimulation events, finger kinematics, contact forces, and millisecond timing) into reliable, interpretable biomarkers of tactile perception and sensorimotor coupling, and then classify normative versus atypical integration patterns with clinical fidelity for assessment.
Expected results: a novel technology and models for the rapid and feasible measurement of tactile-motor deficits in clinical setting, with an initial validation in different neurodevelopmental disorders (i.e. first-episode psychosis, autism spectrum disorder, and dyspraxia). The methods developed and data collected will provide:
(1) an open, versioned feature library for tactile–motor assessment;
(2) classifiers with predefined operating points (sensitivity/specificity);
(3) and an on-device/edge-ready pipeline, i.e. able to run locally on a typical tablet hardware whilst meeting constraints on latency, computing, and data privacy. Success will be measured by reproducibility of features, clinically meaningful effect sizes, and interpretable decision logic that maps back to known neurophysiology rather than artefacts.