Trustworthy & semantic AI
Confidence algebra, provenance, temporal validity, privacy, interoperability, and secure machine knowledge.
Research / Verified record
Research and publications.
This page lists published and accepted research, preprints, historical abstracts and posters, and open-source research artifacts.
15Published peer-reviewed
21Accepted / forthcoming
4Preprints awaiting publication
38Scholarly outputs
Counts overlap: three of the four preprints awaiting publication also have conference acceptances. Earlier manuscripts linked from published records are excluded from the preprint count and filter. The 38 distinct scholarly outputs comprise 15 published peer-reviewed works (including one poster paper), 21 accepted contributions, one standalone preprint, and the published 2018 poster abstract. Historical presentation acceptances and software artifacts appear separately below.
Research topics
Research areas.
The work is grouped by topic.
Agent systems & governance
Context-aware execution, deliberative simulators, controlled agent trees, and auditable recommendation systems.
Edge AI & AIoT
Protocols, compact reasoning, constrained-device MCP, wearable sensing, and embedded intelligence.
Efficient machine learning
Physics-grounded energy metrics, compact architectures, and environmental accountability for compute.
Health & embodied AI
Wearables, medical imaging, speech analysis, privacy-aware biomedicine, and humanoid robotics.
CryptEpi · private researchIMU-JEPA · private research
Compact data systems
Reasoning-carrying binary linked data, hash-augmented structures, compressed-domain processing, and semantic interchange for machines.
Canonical bibliography
Publications & preprints
2026
Beyond Proportional Allocation: Transistor-Level Energy Attribution for Containerized ML Workloads
Accepted · IEEE CloudCom 2026IEEE CloudCom 2026 · Conference paper
Studies transistor-level energy attribution for machine-learning workloads running in containers.
Pollard: Governed Semantic Execution Trees for Framework-Independent AI Agent Control and Replay
Accepted short paper · IEEE ICTAI 2026IEEE ICTAI 2026 · Short paper
Presents governed semantic execution trees for controlling and replaying AI agent work across frameworks.
Hashrope: A Persistent Data Structure for Mechanisms of LLM Context Management
Accepted full paper · IEEE ICTAI 2026IEEE ICTAI 2026 · Full paper
Introduces a persistent data structure for managing large-language-model context.
StegaQR: Learned Private Data Embedding in Standard-Decodable QR Codes
Accepted full paper · IEEE ICTAI 2026IEEE ICTAI 2026 · Full paper
Explores learned private-data embedding in QR codes that remain decodable by standard readers.
Exact Privacy-Preserving Similarity Search via Two-Server Additive Secret Sharing
Accepted full paper · IEEE ICTAI 2026IEEE ICTAI 2026 · Full paper
Examines two-server additive secret sharing for privacy-preserving similarity search.
CBOR-LD-Ex: Compact Linked Data with Reasoning Metadata for Constrained IoT Networks
Accepted regular paper · IEEE WF-IoT 2026IEEE WF-IoT 2026 · Regular paper
Presents compact linked-data encoding with reasoning metadata for constrained IoT networks.
IMU-JEPA: Comparing Latent Prediction and Signal Reconstruction for Self-Supervised Wearable Sensor Representation Learning
Accepted regular paper · IEEE WF-IoT 2026IEEE WF-IoT 2026 · Regular paper
Compares latent prediction and signal reconstruction for self-supervised learning from wearable sensor data.
Zero-Shot Attack-Family Generalization with 1-Bit LLM Routers at the IoT Edge
Accepted regular paper · IEEE WF-IoT 2026IEEE WF-IoT 2026 · Regular paper
Studies how 1-bit language-model routers generalize to unseen attack families at the IoT edge.
Compliance Algebra for IoT Energy Data Governance: Detecting Silent Failures in Smart Meter Pipelines
Accepted regular paper · IEEE WF-IoT 2026IEEE WF-IoT 2026 · Regular paper
Studies compliance algebra for detecting silent governance failures in IoT smart-meter data pipelines.
Drug Integrity Algebra: Quantifying End-to-End Pharmaceutical Safety with Subjective Logic
Accepted main-track paper · IEEE HealthCom 2026IEEE HealthCom 2026 · Main-track paper
Models uncertain pharmaceutical supply-chain evidence with Subjective Logic to assess drug integrity across connected stages.
Privacy-Preserving Confidence-Aware FHIR R4 Exchange: A HIPAA Compliance Algebra Grounded in Subjective Logic
Accepted main-track paper · IEEE HealthCom 2026IEEE HealthCom 2026 · Main-track paper
Proposes uncertainty-aware compliance operators for FHIR R4 exchange and evaluates privacy-preserving aggregation of compliance evidence.
SQT-IoT: A Bandwidth-Efficient Speech Codec for IoT Edge Communications
Accepted · IEEE UEMCON 2026IEEE UEMCON 2026 · Conference paper
Explores a bandwidth-efficient speech codec for communication between IoT edge devices.
Beyond FLOPs: A Vision for Physics-Grounded Energy Intelligence in Machine Learning
Accepted extended abstract · ACM AI Summit 2026ACM AI Summit 2026 · Extended abstract · Poster presentation
Presents a vision for grounding machine-learning energy assessment in physical computation costs beyond operation counts.
corpusgen: An Open-Source Toolkit for Phoneme-Coverage-Optimized Speech Corpus Design Across Languages
Accepted Show & Tell · Interspeech 2026Interspeech 2026 · Show & Tell contribution
Demonstrates an open-source toolkit for designing multilingual speech corpora with optimized phoneme coverage.
Ergonomic Assessment of LLM-Assisted Work among Bangladeshi Technology Users: A Multi-Instrument Survey Study
Accepted · IEEE GHTC 2026IEEE GHTC 2026 · Conference paper
Uses multiple survey instruments to study ergonomic aspects of LLM-assisted work among Bangladeshi technology users.
Semantic Confidence Algebra for Crowdsourced Disaster Report Verification: A JSON-LD Extended Approach
Accepted · IEEE/ACS AICCSA 2026IEEE/ACS AICCSA 2026 · Conference paper
Applies semantic confidence algebra and extended JSON-LD metadata to verification of crowdsourced disaster reports.
Rules Before Oracles: Auditable, User-Configurable Argument Selection for Deliberative Polling
PreprintarXiv:2608.23979 · August 25, 2026
Defines seven audit criteria for civic recommenders and evaluates a user-configurable, one-hop endorsement rule across roughly 17,000 agentic simulations, including adversarial flooding and a peer-to-peer deployment mapping.
Transistor-Level Energy Analysis of Signal Processing Algorithms: Beyond Operation Counting
Accepted · IEEE MLSP 2026IEEE MLSP 2026 · Atlanta, Georgia · September 28–October 1, 2026 · Poster Session 3 scheduled for Thursday, October 1, 1–3 p.m. ET
Extends TOML to 37 signal-processing algorithms across two NVIDIA GPUs, modeling compute, memory, command-dispatch, and fused-sequential costs against profiler and energy measurements.
MultiSpecQR: Multi-Spectral QR Code Encoding and Neural Network-Assisted Decoding via Color Channel Multiplexing
Accepted regular paper · IEEE ICMLA 2026IEEE ICMLA 2026 · Paper presentation scheduled for October 5, 2026, at 11:00 · Special Session 1A (SS1-A)
Combines three independent QR payloads through RGB channel multiplexing and uses neural-assisted decoding, with a palette mode supporting up to nine layers.
TraceCoder: Explainable and Auditable Code Generation with Position-Key Snippet Versioning
Accepted short paper · AGENTICS 2026AGENTICS 2026 · Short paper, poster presentation · Preprint: arXiv:2607.26307 · July 28, 2026
Introduces an auditable code-generation workflow that preserves benchmark-driven repair history, explains how failures shaped individual snippets, and uses stable fractional position keys to track evolving code without disrupting surrounding lines.
Chronofy: A Temporal-Logical Decay Architecture for Information Validity in Time-Aware Retrieval-Augmented Generation
Published full paper · IEEE IRI 20262026 IEEE International Conference on Information Reuse and Integration for Data Science (IRI) · Full paper · pages 454–459 · Earlier manuscript: arXiv:2607.20560
Introduces a three-layer neuro-symbolic architecture that preserves fact age in temporal embeddings, applies learnable decay during retrieval, and verifies the validity of retrieved evidence with Signal Temporal Logic to reduce temporal hallucination and trigger data re-acquisition when context is stale.
Epistemic Edge: Subjective Logic Guardrails for LLM-Driven IoT Actuation
Published short paper · IEEE IRI 20262026 IEEE International Conference on Information Reuse and Integration for Data Science (IRI) · Short paper · pages 37–40
Applies Subjective Logic guardrails to LLM-driven IoT actuation so decisions can account for uncertain, conflicting, or stale edge-state evidence before acting.
The Environmental Cost of Digital Sovereignty: Water, Energy, and Emissions Impacts of Sovereign AI Infrastructure in the Global South
Accepted · IEEE GHTC 2026IEEE GHTC 2026 · Accepted paper · Preprint: arXiv:2607.13443 · July 15, 2026
Models water, electricity, and emissions pressures for sovereign-AI deployment scenarios and proposes principles for environmentally responsible, resource-aware infrastructure.
Evaluation of Alternative-Based Information Systems for Deliberative Polling using an Agentic Simulator
Accepted full paper · AGENTICS 2026AGENTICS 2026 · Accepted full paper · Preprint: arXiv:2606.11692 · June 10, 2026
Introduces an agentic simulator for measuring justification coverage, diversity, and adversarial robustness in large-scale deliberative polling systems.
MCP-Edge: Extending the Model Context Protocol to Constrained IoT Devices
Peer reviewed2026 IEEE Cloud Summit · Washington, DC · June 25–26, 2026 · pages 80–85
Extends the Model Context Protocol to constrained IoT devices through a tiered architecture spanning MCP-Lite endpoints, gateways, and full protocol clients.
JSON-LD 1.2 and Beyond: Extensions for Machine Learning Data Exchange
Peer reviewedThe International FLAIRS Conference Proceedings 39(1) · Marco Island, Florida · May 17–20, 2026
A systematic gap analysis and backward-compatible extension set for representing confidence, provenance, temporal scope, embeddings, integrity, and validation in AI-generated linked data.
TOML: Transistor Operations for Machine Learning: A Physics-Grounded Energy Efficiency Framework
Peer reviewedThe International FLAIRS Conference Proceedings 39(1) · Marco Island, Florida · May 17–20, 2026
Extends transistor-level energy modeling to multiple machine-learning architectures and introduces metrics for switching, memory-compute balance, and capability-normalized efficiency.
Counting Constraints in POMDPs based on PID Controllers
Published poster paperThe International FLAIRS Conference Proceedings 39(1) · Marco Island, Florida · May 17–20, 2026
A Bayesian architecture for incorporating uncertain event-count constraints into POMDP decision processes, integrating probabilistic counts with PID-style control.
Real-Time Neck Posture Classification Using a Lightweight Wearable IMU Pendant
Peer reviewedThe International FLAIRS Conference Proceedings 39(1) · Marco Island, Florida · May 17–20, 2026
Evaluates lightweight models and physics-informed rules for classifying seven neck-posture states from a single wearable IMU on constrained hardware.
2025
Counting Doors by Integrating POMDP Models and PID Controllers
Peer reviewed38th Florida Conference on Recent Advances in Robotics · Full paper · May 9, 2025
Integrates probabilistic door counting into robot planning and control, connecting uncertain perception with POMDP reasoning and PID controllers.
2023
Comparative Evaluation of Transfer Learning for Classification of Brain Tumor Using MRI
Peer reviewed2023 International Conference on Machine Learning and Applications (ICMLA) · December 15, 2023 · pages 697–702
Compares transfer-learning architectures for classifying brain tumors from MRI scans and examines their relative diagnostic performance.
2022
An Insight into the Intricacies of Lingual Paraphrasing Pragmatic Discourse on the Purpose of Synonyms
Journal articleBulletin of Electrical Engineering and Informatics 11(5) · October 1, 2022 · pages 2958–2967
Examines neural paraphrasing and synonym choice with attention to semantic preservation, fluency, and pragmatic discourse.
Multi-Dimensional Spectral Process for Cepstral Feature Engineering & Formant Coding
Journal articleJournal of Electrical Electronics Engineering 1(1) · August 15, 2022 · pages 1–20
Introduces a multi-dimensional speech feature approach and the Speech Quefrency Transform for joint analysis of pitch, formants, and spectral structure.
Medical Imaging a Transfer Learning Process with Multimodal CNN: Dermis-Disorder
Book chapterThird International Conference on Image Processing and Capsule Networks · Lecture Notes in Networks and Systems · July 29, 2022 · pages 556–573
Applies multimodal convolutional transfer learning to the classification of dermatological disorders from medical imagery.
Exploring Public Attitude Towards Children by Leveraging Emoji to Track Out Sentiment Using Distil-BERT a Fine-Tuned Model
Book chapterThird International Conference on Image Processing and Capsule Networks · Lecture Notes in Networks and Systems · July 29, 2022 · pages 332–346
Uses a fine-tuned DistilBERT model and emoji signals to examine public sentiment and online attitudes concerning children.
2019
Towards Enhanced Comprehension of Human Errors in Cybersecurity Attacks
Book chapterAdvances in Human Error, Reliability, Resilience, and Performance · AISC 778 · 2019 · pages 163–175
Studies human error as a factor in cybersecurity incidents and frames improved comprehension as part of more effective defensive systems.
2018
Models and Framework for Supporting Humanoid Robot Planning & Exploration
Conference paper31st Florida Conference on Recent Advances in Robotics · May 10–11, 2018 · pages 1–5
Proposes learned models and a high-level task framework for planning, localization, and exploration with a Nao humanoid robot.
Cooperation Protocols for Ad-Hoc Robot Teams Composition in Labyrinth Exploration
Poster abstractFLAIRS-31 proceedings · AAAI Press · page 552 · May 21–23, 2018
A proceedings poster abstract on coordination protocols for dynamically composing robot teams during labyrinth exploration.
No publications match that filter. Try a different term.
Research artifacts
Software, historical abstracts and presentations.
Software releases, accepted historical presentations, and proposals are separate from the scholarly catalog above.
Current Cybersecurity Threats: Analysis and Solutions From A Human Factors Perspective
Muntaser Syed and Mohammad Algarni. Accepted abstract and program-listed presentation in the Human Factors in Cybersecurity session at AHFE 2016. This record does not claim a published full paper.
Voice controlled autonomous system for hazard detection and rescue operations
Accepted poster for NAECON 2018, with Muntaser Syed registered as presenting author. This historical acceptance is recorded separately from proceedings publications.
Corpusgen v0.1.7
Current archived release of the language-agnostic speech-corpus generator by Muntaser Syed, Marius Silaghi, Sheikh Abujar, Sharun Akter Khushbu, and Fariha Tasmin Jaigirdar.
Voice Controlled Unmanned Aerial Vehicles
Gamal Bohouta, Veton Z. Këpuska, and Muntaser Syed. A presentation on voice interaction and control architectures for UAV systems.
Real-time Data Analysis (Sky Eyes Project)
Gamal Bohouta, Veton Z. Këpuska, Muntaser Syed, and Mahmood Alfathe. A proposed cloud-connected drone platform for live image, video, voice, and sensor analysis.
Source profiles
Publication profiles.
External profiles may contain duplicate or automatically matched entries. Records on this page are checked against primary publication metadata.