SNAI 2024 accepted 24 papers out of 87 submissions, for an overall acceptance rate of 27.6%. The accepted papers were distributed across the four tracks as follows: Track 1 (7 papers), Track 2 (6 papers), Track 3 (6 papers), Track 4 (5 papers). The proceedings are archived in the LiteConf open-access repository.

Papers are listed in the order they appeared in the oral programme. The Best Paper and Best Student Paper awards are marked inline.

Track 1 — AI for Network Management and Optimization (7 papers)

• Best Paper
  Closing the Loop on Learned Traffic Engineering: From Benchmarks to Production
  Liang Zhao (Zhejiang University), Wei Zhang (Tsinghua University), Carlos Mendes (Universidade de São Paulo)
• Robust Learned Congestion Control under Non-Stationary Traffic Mixes
  Hiroshi Tanaka (Kyoto University), Aditi Banerjee (Indian Institute of Science)
• Self-Driving Networks in Practice: A Two-Year Deployment Study at a National Carrier
  Ana Petrova (ETH Zürich), Daniel Okafor (ETH Zürich)
• Intent-Based Network Configuration via Constrained Reinforcement Learning
  Chen Liu (Peking University), Jiang Wei (Shanghai Jiao Tong University)
• Forecasting Optical Transport Utilization with Spatio-Temporal Graph Models
  Sofia Greco (Politecnico di Milano), Marta Ribó (Universitat Politècnica de Catalunya)
• Telemetry Summarization for Root-Cause Analysis in Large Data-Center Fabrics
  Yusuf Demir (Middle East Technical University), Pavel Novak (Czech Technical University in Prague)
• Online Learning for Slice Admission in 5G Core Networks
  Ahmet Yilmaz (Bilkent University), Mohammed Al-Saleh (King Abdullah University of Science and Technology)

Track 2 — Edge and In-Network Inference (6 papers)

• Line-Rate Feature Extraction on Commodity Programmable Switches
  Elena Vasquez (UC San Diego), Diego Alvarez (IMDEA Networks)
• Partitioning DNNs across the Edge and the Data Plane under Latency Budgets
  Rohan Mehta (University of Melbourne), Sarah Whitfield (University of Melbourne)
• SmartNIC Inference at 100 Gbps: A Measurement Study
  Chen Liu (Peking University), Liang Zhao (Zhejiang University)
• Adaptive Resolution for Edge Video Analytics under Bandwidth Constraints
  Priya Subramanian (IIT Madras), Lakshmi Iyer (IIT Bombay)
• Compiler Support for In-Network Operators on Multiple ASIC Targets
  Ana Petrova (ETH Zürich), Pavel Novak (Czech Technical University in Prague)
• Latency-Aware Model Placement for Multi-Tier Edge Networks
  Fatima Zahra (Mohammed VI Polytechnic University), Sofia Greco (Politecnico di Milano)

Track 3 — Distributed and Federated Learning over Networks (6 papers)

• Gradient Compression under Bandwidth-Varying Cellular Links
  Jiang Wei (Shanghai Jiao Tong University), Chen Liu (Peking University)
• Straggler-Tolerant Federated Learning over Asymmetric Residential ISPs
  Marta Ribó (Universitat Politècnica de Catalunya), Maria Costa (INESC-ID Lisboa)
• Decentralized Training over Peer-to-Peer Topologies: A Measurement Study
  Aditi Banerjee (Indian Institute of Science), Rohan Mehta (University of Melbourne)
• Privacy-Preserving Aggregation for Cross-Silo Federated Learning over WANs
  Kenji Nakamura (University of Tokyo), Hiroshi Tanaka (Kyoto University)
• Topology-Aware Scheduling for AllReduce over Lossless Fabrics
  Elena Vasquez (UC San Diego), Diego Alvarez (IMDEA Networks)
• Bandwidth-Aware Quantization for Heterogeneous Federated Clients
  Lakshmi Iyer (IIT Bombay), Priya Subramanian (IIT Madras)

Track 4 — Machine Learning for Network Security (5 papers)

• Best Student Paper
  Adversarial Robustness of Encrypted-Traffic Classifiers under Real-World Distribution Shift
  James O'Connor (Trinity College Dublin), Daniel Okafor (ETH Zürich)
• Long-Tail Detection of Low-Volume DDoS Attacks with Few-Shot Models
  Carlos Mendes (Universidade de São Paulo), Fatima Zahra (Mohammed VI Polytechnic University)
• Operational Deployment of ML-Based Intrusion Detection: A Three-Year Study
  Daniel Okafor (ETH Zürich), Ana Petrova (ETH Zürich)
• Robust Feature Attribution for Network Anomaly Explanation
  Maria Costa (INESC-ID Lisboa), Marta Ribó (Universitat Politècnica de Catalunya)
• Cross-Protocol Attack Attribution with Graph Neural Networks
  Yusuf Demir (Middle East Technical University), Ahmet Yilmaz (Bilkent University)

Awards

• Best Paper — Closing the Loop on Learned Traffic Engineering: From Benchmarks to Production (Track 1)
• Best Student Paper — Adversarial Robustness of Encrypted-Traffic Classifiers under Real-World Distribution Shift (Track 4)
• Distinguished Reviewer Award

Access

The full proceedings, including author-prepared PDFs, are archived at the LiteConf CFP record: https://liteconf.org/cfp/snai-2024/. Authors retained copyright; papers are published under the open-access terms of the symposium.