Live or Lie: Action-Aware Capsule Multiple Instance Learning for Risk Assessment in Live Streaming Platforms

Live streaming has become a cornerstone of today's internet, enabling massive real-time social interactions. However, it faces severe risks arising from sparse, coordinated malicious behaviors among multiple participants, which are often concealed within normal activities and challenging to detect timely and accurately. In this work, we provide a pioneering study on risk assessment in live streaming rooms, characterized by weak supervision where only room-level labels are available. We formulate the task as a Multiple Instance Learning (MIL) problem, treating each room as a bag and defining structured user-timeslot capsules as instances. These capsules represent subsequences of user actions within specific time windows, encapsulating localized behavioral patterns. Based on this formulation, we propose AC-MIL, an Action-aware Capsule MIL framework that models both individual behaviors and group-level coordination patterns. AC-MIL captures multi-granular semantics and behavioral cues through a serial and parallel architecture that jointly encodes temporal dynamics and cross-user dependencies. These signals are integrated for robust room-level risk prediction, while also offering interpretable evidence at the behavior segment level. Extensive experiments on large-scale industrial datasets from Douyin demonstrate that AC-MIL significantly outperforms MIL and sequential baselines, establishing new state-of-the-art performance in room-level risk assessment for live streaming. Moreover, AC-MIL provides capsule-level interpretability, enabling identification of risky behavior segments as actionable evidence for intervention. The project page is available at: https://qiaoyran.github.io/AC-MIL/.

• Publication year

  2026

• Venue

  Unknown venue

• Publication date

  2026-02-03

• Fields of study

  Computer Science

• Identifiers
  DOI 10.1145/3770854.3780246arXiv 2602.03520
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• VLM as Policy: Common-Law Content Moderation Framework for Short Video Platform

  2025cited by this paper
• TAIL-MIL: Time-Aware and Instance-Learnable Multiple Instance Learning for Multivariate Time Series Anomaly Detection

  2025cited by this paper
• Online Fraud Detection via Test-Time Retrieval-Based Representation Enrichment

  2025cited by this paper
• Graph neural networks for financial fraud detection: a review

  2024cited by this paper
• Sequence As Genes: An User Behavior Modeling Framework for Fraud Transaction Detection in E-commerce

  2023cited by this paper
• Detecting rumours with latency guarantees using massive streaming data

  2022cited by this paper
• Additive MIL: Intrinsically Interpretable Multiple Instance Learning for Pathology

  2022cited by this paper
• User Behavior Pre-training for Online Fraud Detection

  2022cited by this paper
• A Holistic Approach to Undesired Content Detection in the Real World

  2022cited by this paper
• AUC-oriented Graph Neural Network for Fraud Detection

  2022cited by this paper
• Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection

  2021cited by this paper
• Intention-aware Heterogeneous Graph Attention Networks for Fraud Transactions Detection

  2021cited by this paper
• Live-Streaming Fraud Detection: A Heterogeneous Graph Neural Network Approach

  2021cited by this paper
• Enhancing Graph Neural Network-based Fraud Detectors against Camouflaged Fraudsters

  2020cited by this paper
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting

  2020cited by this paper
• Fraud Transactions Detection via Behavior Tree with Local Intention Calibration

  2020cited by this paper
• Reformer: The Efficient Transformer

  2020cited by this paper
• Measuring and Characterizing Hate Speech on News Websites

  2020cited by this paper
• Interleaved Sequence RNNs for Fraud Detection

  2020cited by this paper
• Attention is All you Need

  2017influential reference
• Deep multiple instance learning for image classification and auto-annotation

  2015cited by this paper
• On the Properties of Neural Machine Translation: Encoder–Decoder Approaches

  2014cited by this paper
• Locate the Hate: Detecting Tweets against Blacks

  2013cited by this paper
• Solving the Multiple Instance Problem with Axis-Parallel Rectangles

  1997cited by this paper
• Long Short-Term Memory

  1997cited by this paper

• No citing papers are available for this paper.