Chimera: Neuro-Symbolic Attention Primitives for Trustworthy Dataplane Intelligence

TL;DR

Chimera presents a neuro-symbolic attention framework for programmable dataplanes, enabling trustworthy, high-fidelity traffic analysis at line rate within hardware constraints.

Contribution

It introduces a novel mapping of neural attention and symbolic constraints onto dataplane primitives, ensuring predictable and auditable inference.

Findings

Achieves high-fidelity inference within commodity switch resources.

Enables trustworthy, line-rate traffic analysis with symbolic guarantees.

Demonstrates stable operation with a hardware-aware mapping and update scheme.

Abstract

Deploying expressive learning models directly on programmable dataplanes promises line-rate, low-latency traffic analysis but remains hindered by strict hardware constraints and the need for predictable, auditable behavior. Chimera introduces a principled framework that maps attention-oriented neural computations and symbolic constraints onto dataplane primitives, enabling trustworthy inference within the match-action pipeline. Chimera combines a kernelized, linearized attention approximation with a two-layer key-selection hierarchy and a cascade fusion mechanism that enforces hard symbolic guarantees while preserving neural expressivity. The design includes a hardware-aware mapping protocol and a two-timescale update scheme that together permit stable, line-rate operation under realistic dataplane budgets. The paper presents the Chimera architecture, a hardware mapping strategy, and empirical evidence showing that neuro-symbolic attention primitives can achieve high-fidelity inference within the resource envelope of commodity programmable switches.