Halhoul Mhamed M2 in progress
An M2 student developing a path in mathematical modelling, optimization and scientific computing, with interests in inverse problems and learning.
Research interests
Three Master’s scientific projects: restoring an image, studying neural optimization and reconstructing a source from indirect measurements.
M2 in progress
Abdelmalek Essaâdi University · Polydisciplinary Faculty of Larache
01 / Master’s scientific project
Problem, protocol, results and limitations: each study connects to public sources and research artefacts.
Master’s scientific project
Heat versus directional Perona–Malik diffusion on two synthetic images. Noise reduction, edge interpretation and parameter sensitivity, with disagreement between PSNR and SSIM.
Master’s scientific project
Twenty paired initializations compare value-only and derivative-informed supervision. Mean decision-score gains favor M2 within this benchmark; the primary sensitivity-association hypothesis remains inconclusive.
Master’s scientific project
Three controlled studies of learned affine estimators for harmonic boundary reconstruction, with a local covariance benefit and two useful improvements not established.
Documented academic work, not peer-reviewed scientific publications.
02 / Comparative synthesis
A synthesis of three studies examines what the criteria establish, what remains inconclusive and the limits of transferring a finding across scenarios.
Read the study03 / Software project
Two complementary software projects, presented in their published state: a local demo and pre-alpha contracts.
Software project
A software prototype connecting browser microphone input to Gemini Live through WebRTC and an asynchronous Python backend. Its scope is a short French voice demonstration, with no claim of production readiness.
Software project
A pre-alpha software project that separates documentary evidence, unresolved questions and implementation choices in versioned screening rules. The published artifact is a contract, not an operational screening service.
Contact
For conversations with prospective doctoral supervisors, laboratories and scientific partners, please contact me directly.
University email
halhoul.mhamed@etu.uae.ac.maPersonal email
halhoulmhamed@gmail.comResearcher identifier
ORCIDProfessional network
LinkedIn