This work proposes a multimodal anomaly detection framework based on cross-modal reconstruction of heterogeneous time-series sensor data that learns system dynamics by reconstructing each modality from the others, thereby exploiting complementary information across modalities.
Abstract
Fault detection is essential in industrial systems, enabling early identification of abnormal behaviour and improving safety, reliability, and operational efficiency. Modern systems increasingly rely on heterogeneous sensing modalities that capture complementary aspects of the underlying physical process. However, existing data-driven anomaly detection methods often process each modality independently or use simple feature-level fusion, limiting their ability to exploit cross-modal relationships that characterize normal system behaviour. Their performance also commonly assumes similar training and deployment distributions, whereas real-world operation is affected by changing operating conditions, environmental influences, and system degradation that induce distribution shifts and reduce detection performance, especially in unseen regimes. In this work, we propose a multimodal anomaly detection framework based on cross-modal reconstruction of heterogeneous time-series sensor data. Rather than modeling each modality independently, the framework learns system dynamics by reconstructing each modality from the others, thereby exploiting complementary information across modalities. This integrates information across sensing channels without requiring explicit temporal alignment or identical sampling rates, while improving robustness to sensor noise, missing measurements, and modality-specific disturbances. To address distribution shifts during real-world deployment, anomalies are identified using cross-modal reconstruction error and an adaptive test-time thresholding mechanism that adjusts to changing operating conditions. Experiments on three industrial case studies show strong fault detection performance and substantially improved robustness under out-of-distribution conditions, with the largest gains observed in the most challenging operating regimes.
UCTL, an end-to-end unsupervised cross-sensor transfer learning network that enables the transfer and reuse of fault features from labeled source monitoring points to unlabeled target monitoring points with disparate data distributions, thereby achieving cross-sensor domain adaptation in multi-point monitoring systems...
An interpretable multi-sensor fault diagnosis framework, termed LCWC-DAMiT, is proposed that introduces a learnable continuous wavelet convolution (LCWC) module to incorporate time–frequency priors into patch embedding and focuses on physically meaningful fault characteristic frequencies and informative sensor location...
A Kronecker Fused Variational-Evolutionary Reconstruction Framework for Adaptive Intelligent Diagnostics (KROVER-AID), which enables real-time, cross-sensor, and across-sensor fault diagnosis and is a scalable and intelligent approach to predictive maintenance and fault management in next-generation smart manufacturing...
A spectral residual–based evidence fusion network is proposed for industrial fault diagnosis, reducing the influence of randomness, noise, and variable coupling in the frequency domain.
Tong-Kang Zhang, Qian-Fan Xing, Jin-Yang Guan et al.· ASCE-ASME Journal of Risk an...· 0 citations
The proposed multimodal multi‐task learning framework enables simultaneous bearing degradation assessment and remaining useful life prediction by leveraging complementary information from multiple sensing modalities to address the mentioned limitations.
A. N. Sanjrani, Sadiq Ali Shah, N. Q. Soomro et al.· Engineering Reports· 0 citations
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026