arXiv subjects

Joss Armstrong

Publications and source records attributed to Joss Armstrong.

7 recordsLinked to original sources

arXiv2026

Privacy-Preserving Intent Fulfilment and Assurance for 6G RAN

Intent-based network management is the emerging paradigm for 6G service lifecycle automation, with the 3GPP intent management framework (TS~28.312) defining creation, translation, fulfilment, and assurance stages. Existing fulfilment and assurance approaches require deep packet inspection, per-flow state tracking, or access to vendor-internal node telemetry to verify that provisioned resources satisfy expressed intents. These requirements conflict with regulatory constraints (GDPR, ePrivacy Directive) in multi-tenant networks and with vendor opacity in multi-vendor O-RAN deployments. We present an architecture for privacy-preserving intent fulfilment and assurance in which a coordinator provisions resources from declared intent categories without traffic inspection, and verifies fulfilment using only aggregate standardised PM counters at the O1 interface. A data-processing inequality argument shows that the resource allocation reveals at most $\log_2 K$ bits about traffic content, where $K$ is the number of intent categories. We define two architectural privacy properties, intent-traffic unlinkability and node-opaque verification, and show that both hold by construction. Node-opacity does not sacrifice detection power: the aggregate verifier weakly dominates the per-agent verifier under a homogeneity condition. We map the architecture to the 3GPP intent lifecycle and the O-RAN Non-RT RIC, identifying the concrete interfaces, data objects, and deployment points at which the mechanism operates. On production PM counter data from four operator networks, increasing intent-category granularity sharpens provisioning but weakens assurance, consistent with the theoretical prediction that the privacy ceiling is a structural side effect of the detection constraint rather than a separate design parameter.

cs.CR↗

arXiv2026

Designed-Source Reductions and a Dual-Purpose Feasibility Band for Semantic Rate-Distortion

The joint rate-distortion framework of Stavrou and Kountouris (IEEE Transactions on Communications 2023) characterises dual-fidelity tradeoffs for semantic communication on stochastic semantic sources. Many task-oriented communication systems instead use designed sources, where the semantic object is a deterministic oracle allocation $\phi^(t)$ rather than a stochastic quantity given by nature. We isolate the subclass of designed sources under smooth concave utility with assumptions A1, A2 and Euclidean allocation codomain, and restrict the encoder class to deterministic common-category mappings. Within this subclass the SK exponential-tilting decoder and generalised Blahut--Arimoto iteration specialise to conditional-mean decoding and Lloyd--Max stationarity on $\phi^(t)$. When the second fidelity is a monotone single-letter distortion, the joint problem stays inside the SK admissible class; the common-category SK rate is lower-bounded by the max of the corresponding Shannon rate-distortion functions, with equality only when the common-category reconstruction is compatible and RDF-optimal. When the second fidelity is aggregate verification, the joint problem leaves the SK single-letter class and admits a constrained-design feasibility band $R_{\min}(\varepsilon^) \leq R \leq R_{\max}(\beta^)$ of width $\log_2(K_{\max}/K_{\min})$ bits in partition cardinality. The reduction and the band are scope statements on the SK apparatus, not modifications to it. A smart-grid economic-dispatch example with a non-technical-loss-detection contrast illustrates the band.

cs.IT↗

arXiv2026

MARGIN: Runtime Confidence Calibration for Multi-Agent Foundation Model Coordination

Foundation-model pools are increasingly used as black-box responders in coordinated systems where a coordinator must decide which response to trust. Raw self-reported confidence is the natural signal, but is not comparable across models and becomes stale under distribution shift when corrected only at design time. We study runtime confidence calibration for multi-model coordination, where per-model corrections are learned online from deployment outcomes with no model access, no held-out calibration data, and no retraining. Across 18 open-weight foundation models, 8 benchmarks, and over 44,000 observations, we find that online adaptation is a family property: simple same-information online calibrators close most of the calibration gap left by frozen design-time methods under shift, and the forgetting schedule is the dominant design axis. We present MARGIN (Multi-Agent Runtime Grading via Incremental Normalisation), a structured member of this family that maintains per-model, per-confidence-band multiplicative factors using symmetric exponentially weighted updates and shrinkage blending. MARGIN does not dominate the online family on expected calibration error (ECE) under abrupt shift. Its value lies in interpretable confidence-band trust factors, defined cold-start and returning-model behaviour, dynamic-pool support, and a scoped symmetric-update guarantee for fixed-policy non-strategic agents. Empirically, raw verbalized confidence is a weak or misleading pairwise selection signal on hard code-generation tasks, while online calibration substantially improves pairwise resolution and multi-model selection. We also evaluate delayed and selected-answer-only feedback; the latter materially degrades every same-information online method, MARGIN included. Runtime calibration acts as a coordination layer for heterogeneous foundation-model pools, and MARGIN is a practical inspectable instantiation.

cs.LG↗

arXiv2026

BOHM: Zero-Cost Hierarchical Attribution for Compound AI Systems

Compound AI systems route tasks through hierarchies of specialised components. Attribution is dominated by Shapley-based methods (SHAP), which decompose a coalition value function into per-component marginal contributions and require evaluation of the system on arbitrary component subsets. That requirement fails for third-party APIs, opaque endpoints, and agentic orchestrators that concentrate routing on a few tools, leaving most coalitions un-evaluable from the deployed orchestrator. We introduce BOHM, which extracts a hierarchical attribution tree directly from the routing weights such systems already maintain: leaf attribution is the path product of root-to-leaf routing weights; level-k attribution is the induced distribution over depth-k nodes. The method has zero marginal cost, requires no access to component internals, and provides multi-resolution attribution at every level simultaneously, which flat methods cannot offer at any evaluation budget. BOHM and SHAP answer different questions and converge when the deployed router routes near-optimally. On 18 LLMs in a 3-level hierarchy over 880 LiveCodeBench problems, BOHM yields Kendall tau=0.928; SHAP reaches tau=0.980 at 9,000x more coalition evaluations per seed. On a 5-driver, 7-benchmark agentic study (35 cells, complete coverage), drivers concentrate routing on a single tool (top-share median 0.65), and cell-level tau(BOHM,SHAP) is predicted by whether the driver's top pick is the empirically best tool (mean +0.22 vs ~+0.01). On a US Census hierarchy (475 leaves, 4 levels), BOHM recovers ground-truth rankings at every level (tau up to 0.722). BOHM satisfies efficiency, monotonicity, symmetry, and weak suppression but not Shapley's additivity. It is best understood as a complementary primitive: a multi-resolution decomposition computable wherever routing state exists, whose disagreement with Shapley is itself diagnostic.

cs.AI↗

arXiv2026

Implicit Evaluation Under Minimal Information: Hierarchical Component Selection from One-Bit Feedback

A selector that allocates work across opaque components must evaluate them without being told how they performed. We study the least communication that suffices. Each parent maintains an allocation vector over its children and updates it from outcomes by proportional redistribution. Each child reads the sign of its own allocation change, one bit per round it is selected, so no evaluation message crosses the parent-child boundary. (1) The update preserves the simplex with strict positivity. (2) Conditional on being selected, a child's sign is exactly the root outcome at every depth, a sufficient statistic. The conditioning is necessary. (3) Under interiority a single selector has a unique interior equilibrium, exact and closed-form for N=2, with almost sure convergence under decreasing steps. For general N an equi-ratio condition gives an explicit affine equilibrium. (4) The linearised mean flow has real, strictly negative tangent eigenvalues with explicit bounds for all N >= 2. (5) The mean flow converges globally. An explicit potential, up to normalisation the KL divergence between the failure and slack distributions, decreases at a rate equal to the variance of the equi-ratio statistic. Without interiority the rule drops exactly the children below the surviving support's threshold. (6) The update law composes along a single active path, since ancestors select a node's active rounds but do not alter a realised signal. No joint convergence theorem for a hierarchy whose levels adapt simultaneously is claimed. (7) The one-bit interface admits exactly four deterministic memoryless distortions, and at a single selector passing the bit on unchanged delivers strictly higher mean quality upward than negation or constant transmission. Illustrations on synthetic hierarchies up to 16,384 leaves and three real datasets (3.4 billion comparisons, no mismatches) are consistent with the theory.

cs.GT↗

arXiv2026

Source-Side Sufficiency for the Information Bottleneck: Exact Reduction and Finite-Block Equivalence

The input side of the Information Bottleneck may contain task-irrelevant variation that still costs rate. We identify this cost exactly. Let T be a source and C a relevance variable. Suppose the deterministic statistic Z = phi(T) satisfies C-Z-T. For any encoder p(X|T), its conditional average over the fibres of phi preserves I(X;C) and lowers the rate by I(X;T|Z). The reverse pullback preserves both coordinates. These maps establish equality of the relevance-rate curves and Lagrangian infima on standard Borel spaces for every tradeoff parameter. They also characterise all attained optima. Every full-source minimiser factors through Z, and every reduced minimiser pulls back to T. When C is finite and distortion is logarithmic loss, replacement of T^n by Z^n also preserves the optimal remote distortion at every blocklength and message budget. The operational rate-distortion functions are therefore equal.

cs.IT↗

arXiv2026

MISES: Minimal Information Sufficiency for Effective Service

Category-based coordination mechanisms allocate resources by mapping a declared service category to a fixed resource profile, without observing individual demand types. We establish three results for this class of mechanisms. First, the relative welfare gap Delta satisfies a tight two-sided bound in terms of the aggregate within-category allocation variance epsilon: (alpha/2W*)epsilon <= Delta <= (beta/2W*)epsilon. Second, the expected misreporting gain is bounded by the same epsilon without assumptions on agent strategy; demand-derived categories minimise both welfare loss and misreporting incentive simultaneously. Third, aggregate outcome metrics strictly dominate per-agent metrics for service-level detection under a homogeneity condition, for all parameter values, with a finite-sample power gap of O(1/m). At any fixed K, the demand-derived category label is the sufficient statistic for coordination: collecting per-agent data beyond the category label adds noise to the detection problem without reducing the welfare gap. However, welfare and detection impose structurally opposed demands on K: welfare improves with finer categories, detection worsens. The designer faces a feasibility band [Kmin, Kmax] and must choose K within it as a value judgement. We claim that any protocol achieving welfare gap Delta <= epsilon* and missed-detection rate <= beta* requires at least Hlb(epsilon*, beta*) bits of category entropy. We illustrate the mechanism on a synthetic population of 50,000 demand vectors and five weeks of production performance-management data from four anonymised operator networks (28,249 cells).

cs.GT↗