Interpreting Deep Learning-Based Networking Systems

Abstract

While many deep learning (DL)-based networking systems have demonstrated superior performance, the underlying Deep Neural Networks (DNNs) remain blackboxes and stay uninterpretable for network operators. The lack of interpretability makes DL-based networking systems prohibitive to deploy in practice. In this paper, we propose Metis, a framework that provides interpretability for two general categories of networking problems spanning local and global control. Accordingly, Metis introduces two different interpretation methods based on decision tree and hypergraph, where it converts DNN policies to interpretable rule-based controllers and highlight critical components based on analysis over hypergraph. We evaluate Metis over several state-of-the-art DL-based networking systems and show that Metis provides human-readable interpretations while preserving nearly no degradation in performance. We further present four concrete use cases of Metis, showcasing how Metis helps network operators to design, debug, deploy, and ad-hoc adjust DL-based networking systems.

arXiv commentTo appear at ACM SIGCOMM 2020

References in corpus (15)

• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
• The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
• Sensitivity and Generalization in Neural Networks: an Empirical Study
• Unveiling the potential of Graph Neural Networks for network modeling and optimization in SDN
• HyperGCN: A New Method of Training Graph Convolutional Networks on Hypergraphs
• Verifiable Reinforcement Learning via Policy Extraction
• Interpreting and improving natural-language processing (in machines) with natural language-processing (in the brain)
• Fooling Neural Network Interpretations via Adversarial Model Manipulation
• Exploring Interpretable LSTM Neural Networks over Multi-Variable Data
• Playing the lottery with rewards and multiple languages: lottery tickets in RL and NLP
• Programmatically Interpretable Reinforcement Learning
• An Inductive Synthesis Framework for Verifiable Reinforcement Learning
• Real-world Video Adaptation with Reinforcement Learning
• Hyper-SAGNN: a self-attention based graph neural network for hypergraphs
• CoCo: Compact and Optimized Consolidation of Modularized Service Function Chains in NFV

Cited by in corpus (6)

• NetLLM: Adapting Large Language Models for Networking
• DeepAID: Interpreting and Improving Deep Learning-based Anomaly Detection in Security Applications
• Actionable and Interpretable Fault Localization for Recurring Failures in Online Service Systems
• Landing AI on Networks: An equipment vendor viewpoint on Autonomous Driving Networks
• Interpretable Modeling of Deep Reinforcement Learning Driven Scheduling
• TbExplain: A Text-based Explanation Method for Scene Classification Models with the Statistical Prediction Correction