NARCE: A Mamba-Based Neural Algorithmic Reasoner Framework for Online Complex Event Detection

2025-02-11

AI Summary

Real-time online detection of complex events (CEs)—i.e., identifying spatiotemporally constrained sequences of atomic events (AEs) from long-horizon, high-noise sensor streams—remains challenging. Method: We propose Mamba-NAR, a novel collaborative framework that synergistically integrates the state-space model Mamba with neural algorithmic reasoning (NAR), decoupling rule learning from sensor-to-event mapping. To enable zero-shot rule modeling, we leverage large language models (LLMs) to generate conceptual traces; a lightweight adapter then bridges semantic rules to raw sensor signals with high fidelity and efficiency. Results: Extensive experiments demonstrate that Mamba-NAR outperforms state-of-the-art methods across accuracy, cross-type generalization (including unseen CEs and ultra-long sequences), and data efficiency—reducing annotation cost significantly—while maintaining low latency and strong robustness to noise and distribution shifts. This work establishes a new paradigm for real-time CE perception in dynamic sensing environments.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Current machine learning models excel in short-span perception tasks but struggle to derive high-level insights from long-term observation, a capability central to understanding complex events (CEs). CEs, defined as sequences of short-term atomic events (AEs) governed by spatiotemporal rules, are challenging to detect online due to the need to extract meaningful patterns from long and noisy sensor data while ignoring irrelevant events. We hypothesize that state-based methods are well-suited for CE detection, as they capture event progression through state transitions without requiring long-term memory. Baseline experiments validate this, demonstrating that the state-space model Mamba outperforms existing architectures. However, Mamba's reliance on extensive labeled data, which are difficult to obtain, motivates our second hypothesis: decoupling CE rule learning from noisy sensor data can reduce data requirements. To address this, we propose NARCE, a framework that combines Neural Algorithmic Reasoning (NAR) to split the task into two components: (i) learning CE rules independently of sensor data using synthetic concept traces generated by LLMs and (ii) mapping sensor inputs to these rules via an adapter. Our results show that NARCE outperforms baselines in accuracy, generalization to unseen and longer sensor data, and data efficiency, significantly reducing annotation costs while advancing robust CE detection.

Problem

• Detects complex events from long-term noisy data.
• Reduces dependency on extensive labeled data.
• Improves accuracy and generalization in event detection.

Innovation

• Mamba-based state-space model
• Neural Algorithmic Reasoning framework
• LLMs for synthetic concept traces