arXiv subjects

Minhao Jin

Publications and source records attributed to Minhao Jin.

5 recordsLinked to original sources

arXiv2026

The Frontier LLM Trap in Network Automation

Large LLMs are powerful tools for network automation, but they are expensive, slow to serve, hard to audit, poorly tailored to individual networks, and create long-term dependencies on a small number of AI providers. Existing alternatives fall short: small open-source models are cheaper but unreliable, while deterministic scripts and verification are controllable but hard to build and maintain. We propose a middle ground. An offline loop composed of fuzzing and validation discovers the recurring mistakes small models make, then resulting networking knowledge is expressed as explicit logic rules. In production, these rules guide a small model on tasks such as configuration translation, yielding automation that is cheaper, lower-latency, auditable, and easier to adapt to a specific network. More critically, network knowledge and operational experience stay and are accumulated where they belong, the network itself, not a rented service.

cs.NI↗

arXiv2026

Worst-Case Discovery and Runtime Protection for RL-Based Network Controllers

RL-based controllers achieve strong average-case performance in networking tasks such as congestion control and adaptive bitrate streaming. Yet their performance can degrade severely under network conditions where strong performance is still achievable. Identifying such conditions and quantifying the resulting performance gap is intractable by enumeration, while the sequential and closed-loop nature of RL controllers makes formal verification methods impractical. We present ReGuard, a framework that discovers worst-case scenarios for a given RL controller and protects it against them at inference time without retraining. Discovery is formulated as a bilevel regret-maximization problem, which yields a certified lower bound on the worst-case performance gap. The discovered trajectories are then analyzed as counterfactuals and compiled into lightweight logic rules that intervene only when a risky state is detected, leaving the controller's behavior unchanged otherwise. We evaluate ReGuard across three RL-based network controllers: Pensieve, Sage, and Park. ReGuard discovers scenarios in which the controller's performance is 43$-$64% worse than what is achievable. ReGuard not only discovers gaps 57% to 6$\times$ larger than those found by the strongest baselines but also shrinks them by 79$-$85% via lightweight rule-based protection while preserving nominal performance. ReGuard's protection extends beyond the scenarios it discovers, improving performance across a wider range of network conditions.

cs.NI↗

arXiv2025

Cross-Flow Correlations Survive Synthesis: Measuring Source-Level Privacy Leakage in Synthetic Network Traces

Synthetic network data generators (SynNetGens) are increasingly used to share realistic traffic traces without exposing sensitive raw data. While substantial effort has gone into improving fidelity, privacy is either assumed to be a built-in property of synthesis or addressed through differential privacy at the packet or flow level. This paper uncovers a fundamental privacy vulnerability: SynNetGens preserve cross-flow behavioral correlations that expose source-level membership, allowing an attacker to determine whether traffic of specific user, or service was included in the training data. This leakage arises from a mismatch in abstraction: existing SynNetGens operate and are protected at the packet or flow level, while sensitive information is encoded in correlations across flows from the same source. To demonstrate that this vulnerability is exploitable in practice, we develop TraceBleed, the first source-level membership inference attack against black-box SynNetGens. Our evaluation spans five datasets and six SynNetGens, revealing that: (i) every generator leaks source-level information on at least some datasets; (ii) flow- or packet-level differential privacy fails to protect source privacy unless fidelity is degraded to unusable levels; and (iii) releasing 10X more synthetic data amplifies leakage by 130% on average. To support ongoing research in this area, we will maintain a public privacy-fidelity leaderboard so practitioners can choose generators that fit their needs and researchers can benchmark new designs faithfully.

cs.CR↗

arXiv2025

Making Logic a First-Class Citizen in Generative ML for Networking

Generative ML models are increasingly popular in networking for tasks such as telemetry imputation, prediction, and synthetic trace generation. Despite their capabilities, they suffer from two shortcomings: \emph{(i)} their output is often visibly violating well-known networking rules, which undermines their trustworthiness; and \emph{(ii)} they are difficult to control, frequently requiring retraining even for minor changes. To address these limitations and unlock the benefits of generative models for networking, we propose a new paradigm for integrating explicit network knowledge, in the form of first-order logic rules, into ML models used for networking tasks. Rules capture well-known relationships among observed signals, e.g., that increased latency precedes packet loss. While the idea is conceptually straightforward, its realization is challenging: networking knowledge is rarely formalized into rules, and naively injecting rules into ML models often hampers their effectiveness. This paper introduces NetNomos, a multi-stage framework that \emph{(i)} learns rules directly from data (e.g., measurements); \emph{(ii)} filters them to select semantically meaningful ones; and \emph{(iii)} enforces them through collaborative generation between an ML model and a Satisfiability Modulo Theories (SMT) solver. %We evaluate NetNomos both component-wise and end-to-end across four diverse network datasets. We show that NetNomos learns diverse, meaningful rules from four real-world datasets and is 1.6--6.5$\times$ more scalable than DuoAI, a state-of-the-art (SOTA) rule-learning method. By enforcing these rules on a generic GPT-2 model, NetNomos achieves performance on par with or even surpassing specialized SOTA systems such as Zoom2Net and NetShare across three networking tasks: telemetry imputation, traffic forecasting, and synthetic data generation.

cs.NI↗

arXiv2024

Robustifying ML-powered Network Classifiers with PANTS

Multiple network management tasks, from resource allocation to intrusion detection, rely on some form of ML-based network traffic classification (MNC). Despite their potential, MNCs are vulnerable to adversarial inputs, which can lead to outages, poor decision-making, and security violations, among other issues. The goal of this paper is to help network operators assess and enhance the robustness of their MNC against adversarial inputs. The most critical step for this is generating inputs that can fool the MNC while being realizable under various threat models. Compared to other ML models, finding adversarial inputs against MNCs is more challenging due to the existence of non-differentiable components e.g., traffic engineering and the need to constrain inputs to preserve semantics and ensure reliability. These factors prevent the direct use of well-established gradient-based methods developed in adversarial ML (AML). To address these challenges, we introduce PANTS, a practical white-box framework that uniquely integrates AML techniques with Satisfiability Modulo Theories (SMT) solvers to generate adversarial inputs for MNCs. We also embed PANTS into an iterative adversarial training process that enhances the robustness of MNCs against adversarial inputs. PANTS is 70% and 2x more likely in median to find adversarial inputs against target MNCs compared to state-of-the-art baselines, namely Amoeba and BAP. PANTS improves the robustness of the target MNCs by 52.7% (even against attackers outside of what is considered during robustification) without sacrificing their accuracy.

cs.CR↗