arXiv subjects

Kyandoghere Kyamakya

Publications and source records attributed to Kyandoghere Kyamakya.

6 recordsLinked to original sources

arXiv2026

Policy-as-Skill: Governed LLM Decision Support with Evidence, Deterministic Control, and Audit

Organizations increasingly use LLMs for policy, compliance, risk, and operational decision support, requiring evidence validation, review routing, version control, and auditability. We introduce Policy-as-Skill (PaS), a modular runtime that packages these functions as executable, versioned policy capabilities. Thirteen methods are evaluated with a fixed Gemma4 backend on 600 development tasks. PaS+Audit achieves 53.8% exact accuracy, macro-F1 0.346, review F1 0.854, citation precision 1.000, policy-reference recall 0.984, and audit completeness 1.000, outperforming LLM+RAG on most governance and review metrics. Deterministic control raises aggregate accuracy to 61.2% but is strongly task dependent, supporting selective rather than universal rule-based intervention.

cs.AI↗

arXiv2026

TinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers

Replacing attention in a pretrained language model is a compatibility problem: a plausible substitute may alter representations expected by later layers. TinyCeNN-LM introduces a \emph{quality-gated post-training conversion} framework using CeNN-inspired cellular-recurrent layers with bounded local processing, compact recurrent memory, routing, fusion, and accept-or-rollback validation. Three implementations are studied: Integrated Memory, MemoryFusion, and PDelta3-GDN2-CLVR+Local32. Strict PDelta3 conversion accepts a layer only when representation and NLL criteria pass fixed thresholds. On SmolLM2-135M, layers 0-2 are accepted with cumulative $Δ\mathrm{NLL}=+0.01209$, while layer 3 is rejected despite acceptable NLL because representation fidelity fails. On Qwen3.5-0.8B, full-attention layers 3, 7, and 11 are accepted with final $Δ\mathrm{NLL}=+0.02073$. Integrated Memory keeps perplexity within $-0.07\%$ to $+0.93\%$ while reducing total cache by up to $6.01\%$. A sampled 200-item downstream sanity check gives $28.5\%$--$32.0\%$ overall accuracy for converted Qwen releases. The results support conservative, quality-gated structural conversion rather than universal attention replacement or speedup.

cs.AI↗

arXiv2026

$PG-NODE^{TB}$: Physics-Guided Neural Ordinary Differential Equations for Tuberculosis Transmission Dynamics

Tuberculosis (TB) remains a leading global infectious disease, causing approximately 1.3 million deaths and 10.6 million new infections annually. Classical compartmental ODE models are the standard epidemiological tool for TB, yet their fixed-parameter structure cannot adapt to time-varying dynamics, unmodeled effects, or heterogeneous real-world data. This paper presents a methodological framework and proof-of-concept for applying Physics-Guided Neural Ordinary Differential Equations (PG-NODE) to TB transmission modeling within a SLIR (Susceptible, Latent, Infectious, Recovered) compartmental framework. We perform a rigorous mathematical analysis of the SLIR model, including derivation of the basic reproduction number $\mathcal{R}_0$, equilibrium analysis, and normalized sensitivity indices. We then reformulate the SLIR system as a PG-NODE, preserving compartmental conservation laws and biological constraints while enabling neural network components to learn unknown or time-varying rate functions from data. Three simulation scenarios illustrate the framework's intended capabilities: (i) adaptive tracking of time-varying transmission rates, (ii) correcting for unmodeled treatment and relapse dynamics with 27\% lower RMSE than the classical SLIR, and (iii) comparative forecasting of competing intervention policies over a 20-year horizon. Simulation results indicate that PG-NODE has strong potential for improving predictive accuracy while maintaining epidemiological interpretability; full adjoint-based training on real WHO surveillance data is identified as the key next step for empirical validation.

math.DS↗

arXiv2024

Amplitude of Random Telegraph Noise in Junctionless FinFET with Different Channel Shape

The influence of the channel shape in a junctionless silicon-on-insulator finned field-effect transistor (JL SOI FinFET) on the amplitude of random telegraph noise (RTN) induced by single interface trapped charge has been simulated for the transistors with rectangular, trapezoidal, and triangular fin cross sections. The simulation of the RTN amplitude distribution along the channel induced by a single charge trapped at interface defect located at the fin top and at sidewall of JL SOI FinFETs with channels of different shapes is considered. It is established that at trapping the single charge at sidewall surface of the channel, the lowest RTN amplitude is seen for the triangular cross-section and the highest for the rectangular and trapezoidal cross-sections. At the single charge trapping at the top surface of the channel, the RTN amplitude is higher for the rectangle than for the trapezoidal cross-section.

cond-mat.mes-hall↗

arXiv2022

Modelling DDoS Attacks in IoT Networks using Machine Learning

In current Internet-of-Things (IoT) deployments, a mix of traditional IP networking and IoT specific protocols, both relying on the TCP protocol, can be used to transport data from a source to a destination. Therefore, TCP-specific attacks, such as the Distributed Denial of Service (DDoS) using the TCP SYN attack, are one of the most plausible tools that attackers can use on Cyber-Physical Systems (CPS). This may be done by launching an attack from its IoT subsystem, here referred to as the "CPS-IoT", with potential propagation to the different servers located in both fog and the cloud infrastructures of the CPS. This study compares the effectiveness of supervised, unsupervised, and semi-supervised machine learning algorithms for detecting DDoS attacks in CPS-IoT, particularly during data transmission to and from the physical space to the cyber space via the Internet. The algorithms considered are broadly grouped into two: i) Detection algorithms, which include Logistic Regression (LGR), K-Means, and Artificial Neural Networks (ANN). We also looked into the effectiveness of semi-supervised hybrid learning models, which use unsupervised K-Means to label data, then feed the output to a supervised learning model for attack detection. ii.) Prediction algorithms - LGR, Kernel Ridge Regression (KRR) and Support Vector Regression (SVR), which were used to predict imminent attacks. Experimental tests were carried out and obtained results showed that the hybrid model was able to achieve 100% accuracy with zero false positives; while all the prediction models were able to achieve over 94% attack prediction accuracy.

cs.DC↗

arXiv2009

A Differential Cryptanalysis of Yen-Chen-Wu Multimedia Cryptography System (MCS)

At ISCAS'2005, Yen et al. presented a new chaos-based cryptosystem for multimedia transmission named "Multimedia Cryptography System" (MCS). No cryptanalytic results have been reported so far. This paper presents a differential attack to break MCS, which requires only seven chosen plaintexts. The complexity of the attack is O(N), where $N$ is the size of plaintext. Experimental results are also given to show the real performance of the proposed attack.

cs.CR↗