Skip to content
Dr. Jeff Owino

Research

Researching Intelligence. Building Impact.

A research practice concerned with models that generalise, remain private where they must, and travel from the paper into clinics, farms and institutions.

Areas

Research areas

Primary lines of inquiry. Several overlap by design. Audio intelligence is also federated learning; computer vision is also social impact.

01

Machine Learning

Supervised, unsupervised and adaptive learning systems for structured and unstructured data.

02

Deep Learning

Neural architectures for representation learning, sequence modelling and multimodal intelligence.

03

Computer Vision

Visual recognition for agriculture, public health screening and environmental monitoring.

04

Predictive Analytics

Forecasting, risk modelling and decision support built on statistical and machine-learning methods.

05

Statistical Learning

Inference, uncertainty and evaluable models, the statistical backbone of applied AI.

06

Spatial Analytics

Geographic and spatial statistical methods for supply chains, health and environmental systems.

07

Audio Intelligence

Paralinguistic audio analysis, cry classification and privacy-preserving speech processing.

08

AI for Social Impact

Applied intelligence for healthcare, agriculture, climate and communities where expertise is scarce.

Programmes

Featured research

Longer entries for the scholarly projects that define the lab, from infant cry intelligence to crop surveillance, drug design and document digitisation.

FIGURE 01 / AUDIO INTELLIGENCE

f (kHz) · t (s)

RESEARCH / 01 · Healthcare · Deep Learning · Audio AI

Infant Cry Intelligence Programme

Spanning four scholarly projects and multiple publications: adaptive bandit CRNNs, domain-agnostic causal-aware transformers, privacy-enhancing federated learning with denoising regularisation, and an artificial-parenting Android application for autonomous cry surveillance.

A programme of deep-learning audio analytics for classifying infant cry paralinguistics, from attentive CRNNs and bandit modality selection to federated transformers and causal-aware audio models.

Open related work
FIGURE 02 / COMPUTER VISION
leaf · 0.91
lesion · 0.76

RESEARCH / 02 · Agriculture · Computer Vision · Mobile

Autonomous Crop Disease Surveillance

Visual recognition systems designed to detect and monitor crop disease in field conditions, translating computer vision research into a mobile decision-support tool for farmers and extension officers.

AI-powered crop disease detection and monitoring using image processing and machine learning, deployed as an Android application for field use.

Open related work
FIGURE 03 / COMPUTATIONAL BIOLOGY

G = (V, E) · VP35

RESEARCH / 03 · Healthcare · Computational Biology

Computer-Aided Drug Design for Ebola VP35

Computational intelligence applied to viral protein drug discovery. The work placed third at KAPS 2019 and progressed to a commercial outcome with San Francisco Pharmaceuticals.

Machine-learning approaches to computer-aided design of candidate compounds targeting Ebola VP35, recognised at the KAPS 2019 hackathon and advanced toward commercialisation.

Open related work
FIGURE 04 / DOCUMENT INTELLIGENCE
leaf · 0.91
lesion · 0.76

RESEARCH / 04 · Computer Vision · Document AI · Mobile

Records Digitization with Computer Vision

A document intelligence system that captures, classifies and digitises manual record workflows, reducing transcription burden and improving data availability for downstream analytics.

Automation of manual paper records into electronic records using convolutional neural networks, deployed as both an Android application and a web-based platform.

Open related work

Archive

Scholarly projects

CV-indexed research projects mapped to publications and applied work case studies.

  • 01

    Adaptive infant cry classification using multi-armed bandit modality selection in an attentive convolutional recurrent neural network model

    Complex & Intelligent Systems, 2025

    PublicationCase study

  • 02

    Domain-Agnostic Causal-Aware Audio Transformer for Infant Cry Classification

    IC2IE, 2025

    PaperCase study

  • 03

    Autonomous Surveillance of Infants' cries using Deep Learning Audio analytics model (artificial parenting Android app)

    Journal of Data Analysis and Information Processing, 2022

    PublicationCase study

  • 04

    Privacy-Enhancing Infant Cry Classification with Federated Transformers and Denoising Regularization

    International Conference on Computer Engineering, Network and Digital Technologies, 2025

    PaperCase study

  • 05

    Autonomous Surveillance of Crops disease using AI machine learning Technique (Image processing Technique), deployed as Android application

    Case study

  • 06

    Computer Aided drug design for Ebola VP35

    Case study

  • 07

    Automation of manual data records to electronic records using computer vision algorithm (CNN), deployed as Android app and web application

    Case study

Literature

Publications

Filter and search the current record. Sample entries are labelled as placeholders; IEEE and arXiv papers are drawn from public sources.

View Google Scholar Profile

10 records

  • 2026

    Federated Primitive-Preserving Audio Transformers for Non-Identifiable Infant Cry Classification

    Geofrey Owino, Bernard Shibwabo Kasamani, Ahmed M. Abdelmoniem, Edem Wornyo

    IEEE Access

    Formalises federated infant cry classification as a task-non-identifiable learning problem and proposes a Primitive-Preserving Audio Transformer that decouples transferable paralinguistic primitives from client-specific decision boundaries.

    0 citationsRead PaperDOIScholar
    • Artificial Intelligence
    • Machine Learning
    • Healthcare

  • 2026

    Federated Causal-Aware Audio Transformer for Cross-Site Infant Cry Paralinguistic Classification

    Geofrey Owino, Bernard Shibwabo Kasamani, Ahmed M. Abdelmoniem, Edem Wornyo

    ICNLP

    A federated causal-aware audio transformer for cross-site infant cry paralinguistic classification.

    0 citationsScholar
    • Artificial Intelligence
    • Machine Learning
    • Healthcare

  • 2026

    Adaptive Decision-Making in Audio Classification: A Systematic Review of Reinforcement Learning and Multi-Armed Bandit Frameworks

    Geofrey Owino, Timothy Kamanu, John Ndiritu

    Machine Learning and Knowledge Extraction

    A systematic review of reinforcement learning and multi-armed bandit frameworks for adaptive decision-making in audio classification.

    0 citationsScholar
    • Machine Learning
    • Other

  • 2025

    Adaptive Infant Cry Classification Using Multi-Armed Bandit Modality Selection in an Attentive Convolutional Recurrent Neural Network Model

    Geofrey Owino, Timothy Kamanu, John Ndiritu, Conlet Biketi Kikechi

    Complex & Intelligent Systems

    An attentive convolutional-recurrent architecture with multi-armed bandit modality selection for adaptive infant cry classification.

    5 citationsScholar
    • Machine Learning
    • Healthcare
    • Statistics

  • 2025

    Advances in Infant Cry Paralinguistic Classification: Methods, Implementation, and Applications: Systematic Review

    Geofrey Owino, Bernard Shibwabo

    JMIR Rehabilitation and Assistive Technologies

    A systematic review of methods, implementation patterns and applications in infant cry paralinguistic classification, screening more than 5,000 records across nine databases.

    5 citationsScholar
    • Healthcare
    • Machine Learning
    • Other

  • 2025

    Noise-Resilient Bioacoustics Feature Extraction Methods and Their Implications on Audio Classification Performance: Systematic Review

    Geofrey Owino, Bernard Shibwabo

    JMIR Biomedical Engineering

    A systematic review of noise-resilient bioacoustics feature extraction methods and their implications for audio classification performance.

    4 citationsScholar
    • Machine Learning
    • Healthcare
    • Other

  • 2025

    A Federated Multi-Task, Multi-View Metatransformer with Causal-Acoustic Attention for Infant Cry Classification

    Geofrey Owino, Bernard Shibwabo Kasamani, Edem Wornyo

    ICRAAI

    A federated multi-task, multi-view metatransformer with causal-acoustic attention for infant cry classification.

    0 citationsScholar
    • Artificial Intelligence
    • Machine Learning
    • Healthcare

  • 2025

    Privacy-Enhancing Infant Cry Classification with Federated Transformers and Denoising Regularization

    Geofrey Owino, Bernard Shibwabo

    International Conference on Computer Engineering, Network and Digital Technologies

    Investigates federated transformer architectures with denoising regularisation for infant cry classification under privacy constraints.

    0 citationsRead PaperScholar
    • Artificial Intelligence
    • Machine Learning
    • Healthcare

  • 2025

    Domain-Agnostic Causal-Aware Audio Transformer for Infant Cry Classification

    Geofrey Owino, Bernard Shibwabo Kasamani, Ahmed M. Abdelmoniem, Edem Wornyo

    IC2IE

    Introduces DACH-TIC, a hierarchical audio transformer that integrates causal reasoning, multi-task modelling and adversarial domain generalisation for robust infant cry classification under acoustic perturbation.

    0 citationsRead PaperScholar
    • Artificial Intelligence
    • Machine Learning
    • Healthcare

  • 2022

    Autonomous Surveillance of Infants' Needs Using CNN Model for Audio Cry Classification

    Geofrey Owino, A. Waititu, A. Wanjoya, J. Okwiri

    Journal of Data Analysis and Information Processing

    Deep-learning audio classification for autonomous surveillance of infant needs from cry signals.

    1 citationScholar
    • Machine Learning
    • Healthcare
    • Artificial Intelligence