💻 Category

cs

2661 papers

💻 computer science

Policy-Gated Zero-Trust Federated Learning: Identity-Bound Enrollment, Replay-Resistant Control, and Secure Coordination

This paper proposes and validates a policy-gated zero-trust architecture for federated learning that decouples network authentication from model influence by binding identities to signed tokens, enforcing strict enrollment leases with cryptographic nonces, and rigorously testing against 22 adversarial scenarios to ensure only authorized, fresh, and non-revoked participants can influence the global model.

Md Shahanur Islam Shagor2026-09-10
💻 computer science

Adaptive Recursive Learning in Mixture of Experts: A Systematic Review

This systematic review distinguishes between adaptive routing and recursive learning in Mixture-of-Experts (MoE) architectures to propose the Recursive Adaptive Recursive Learning (RARL) framework, which synthesizes current literature on dynamic computation and expert specialization while outlining a research agenda to address challenges like routing stability and resource-constrained intelligent allocation.

SIMAR SINGH RAYAT2026-09-10
💻 computer science

Project-Aware Validation in Software Defect Prediction: A Controlled Simulation and Real-World Benchmark Study of Evaluation Optimism

This study demonstrates through controlled simulations and a secondary analysis of real-world benchmarks that using pooled random train/test splits in software defect prediction yields systematically optimistic performance estimates compared to project-aware validation methods, highlighting the critical need for evaluation protocols that respect project boundaries.

Vladimir Tomilov2026-09-10
💻 computer science

Quorum-Bounded Asynchronous Federated Learning under Non-IID Data and Adversarial Clients: A Systems Study of Stale-Update Exclusion and Convergence

This study demonstrates that a quorum-bounded asynchronous federated learning design effectively reduces straggler-induced latency by up to 72% without significantly compromising final accuracy, though it reveals that data heterogeneity critically amplifies the convergence damage caused by adversarial clients.

Md Shahanur Islam Shagor2026-09-10
💻 computer science

A Severity-Calibrated Adversarial Benchmark for Federated Learning: From Label Corruption to Structured Model-Update Injection

This paper introduces a severity-calibrated adversarial benchmark for federated learning that evaluates five distinct attack families across varied conditions, demonstrating that attack magnitude alone is an insufficient indicator of damage and that robust security evaluation must account for attack geometry, malicious population share, and aggregation rules.

Md Shahanur Islam Shagor2026-09-10
💻 computer science

How Far Can a Single Vector Carry a Language? Mechanistic Limits of Inference-Time Steering for Low-Resource Devanagari Languages

This paper investigates the mechanistic limits of inference-time steering to generate low-resource Devanagari languages (Maithili, Nepali, and Bhojpuri) using Hindi representations in large multilingual models, finding that while a single vector shift can induce target language adherence, it inevitably causes fluency collapse, revealing that steerability is constrained by linear separability rather than model competence or vocabulary overlap.

Sumit Yadav, Santosh Giri, Ganesh Gautam2026-09-10
💻 computer science

Trusting the Null: Positive-Control Fuzzing of Post-Quantum Decode and Verify Paths in Java

This paper demonstrates that while a large-scale fuzzing campaign found no defects in Bouncy Castle 1.85's post-quantum decode and verify paths, a positive-control approach successfully validated the testing harness by uncovering critical length-validation flaws in both Bouncy Castle 1.84 and multiple JDK releases, revealing that these implementations differ only in the timing of their checks rather than the logic of their decisions.

Arpan Sharma2026-09-10
💻 computer science

Drosophila Connectome Topology Provides No Measurable Language- Model Advantage: A Preregistered Negative Experimental Program

This preregistered study systematically tested whether Drosophila brain topology could improve language models and found that, despite various architectural integrations, fly-derived wiring provided no measurable advantage over random, degree-matched, or vanilla baselines in reducing negative log-likelihood or improving associative retrieval.

Vladislav Matyukhin2026-09-10
💻 computer science

Latent Attribution Regularization: Cross-Domain Attribution-Consistency Training for Stable NLP Explanations

This paper introduces Latent Attribution Regularization (LAR), a training-time method that enforces attribution consistency under semantically preserving synonym substitutions, demonstrating that it significantly improves the stability of gradient-based explanations in fine-tuned Transformers without compromising model accuracy, while also establishing that proper token alignment is essential for accurately measuring such stability.

Junaid Hassan2026-09-10