Research interests: Biostatistics – Applied statistics – Forest modeling – Pharmacometrics – Agroforestry
- Learning methods such as Artificial Neuronal networks
- Parametric and nonparametric nonlinear mixed effect model
- Modeling and simulation of biologic phenomena
- Pharmacometrics, Epidemiology and Biometrics
- Quantitative Ecology, Climate change, Agronomic and Environment
After the baccalaureate, I did 5 years of study at the Polytechnique School of Abomey-Calavi (EPAC / UAC) where I graduated as a conception engineer in environmental engineering. I also doubled this degree from a Bachelor’s Degree in Psychology of Social and Professional Life at the Faculty of Arts and Human Sciences (FLASH / UAC).
After three years of professional life, having coordinated two environmental, health and social projects; cumulated to the role of research assistant that I was playing in applied statistics at the university. So I decided to specialize in the field. This led me to do a master of research in statistics, major biostatistics at the Laboratory of Biomathematics and Forestry Estimates (LaBEF / UAC) for two years. Passionate about this field, I had the exchange to register as a doctoral thesis in Mathematical Statistics and Probability at the Institute of Mathematics and Physical Sciences (IMSP / UAC), where I am currently conducting my work on optimization neural network of type multilayer perceptrons.
Modeling empirical data for forecasting purposes is a difficult statistical task because of the relationships that underlie the measurements that can be strongly non-linear, non- univocal, noisy, and of dynamic nature. For this purpose, the standard statistical methods used such as regression models, time series for the treatment of these types of data give mostly unsatisfactory results. It is therefore necessary to explore alternative approaches to develop better models. Thus, new tools and methodologies have been increasingly developed and are able to process increasingly complex data. These include: boltz machine, machine-supported vectors, artificial neural networks (ANN). These are inspired by neurobiology and perform calculations similar to those of the human brain. Attractiveness for ANNs comes from the processing of characteristic information such as nonlinearity, parallelism, noise tolerance, learning ability and of generalization. They are applied in various fields including the field of agriculture, environment, industry, medicine, communication, computer science, electronics, mechanics, insurance automotive, chemistry, biology, air transport, automatic etc. They encompass several types of models, among of them the Multilayer Perceptrons (MCPs), have demonstrated their effectiveness in prevision and prediction of empirical data. Often, these data have a high proportion of contaminated observations with errors whose magnitude and structure can be arbitrary. Thus, the problem of global optimization of learning algorithms used in RNA arises. In addition, although neural networks have demonstrated their ability to obtain excellent predictive and estimated performance, the problem of the choice of hyper-parameters is very critical for these networks because the size of the search space increases significantly of exponential way with the number of intermediate layers. The objectives of this thesis are to: (i) define new models and learning algorithms for PMCs that correct the explanatory variables and explain « at best » so as to find the « real » relationship that binds them; (ii) determine the optimal structure of the PMCs; (iii) analyze their behavior in the face of missing data. Then we will end up with applications for crop yield forecasts
Awards and distinctions
- 2007-2009. Scholarship : Benin government
- 2016-2019. Wold Bank scholarship
- 2010. Fellowship: Inter-University Project Targeted (PIC)/Teck spinneret, Benin-belgium cooperation
- 2009. Prize of excellence: Association of Environmentalists, Benin
- 2009-2010.Small grant: Help Fund of Environmental Protection, United Nations Development Program, Benin
Profiles and Curriculum
Total number of publications
- Articles scientifiques publiés dans les journaux: 7
- Communications: 3
- Documents techniques: 1
- Savi, M.K.,Tchandao Mangamana E.,Deguenon J.M., Hounmenou, C.G. and Glèlèn Kakaï, R. (2017) Determination of Lethal Concentrations Using an R Software Function Integrating the Abbott Correction. Journal of Agricultural Science and Technology A(7):25-30.
- KATE S., HOUNMENOU C. G., AGBANGBA C. E., DEGUENON D. D. S., Gbaguidi M., NAKOU L. G. K. et SINSIN B., 2017. Effets de l’indice de température et d’humidité relative de l’air sur la fécondité des bovins en zone agropastorale de Banikoara (Nord-Bénin). e-Journal of Science & Technology, 12 (3). 31-48 pp.
- KATE, G. C. HOUNMENOU., A. AMAGNIDE., P. V. HOUNDONOUGBO., A. TCHOBO., B. TENTE., S. DIARRA et B. SINSIN. 2015. Changement climatique, mécanisme actuel de prévention et de Gestion des conflits entre agriculteurs et éleveurs en zone agro Pastorale de production cotonnière au Nord Bénin : Cas de la commune de Banikoara. African Crop Science Journal, 23 (1) 9-26. Impact Factor; 2.486
- Sabaï KATE, Castro G. HOUNMENOU, Marc Aurèle CHABI ADJOBO et Brice SINSIN., 2015. Effets des changements climatiques sur les activités agricoles dans commune de Banikoara (NordBénin). Revue Agro-Ecologie 01 (01) 33-43.
- Sabai KATE, Aubin AMAGNIDE, Castro G. HOUNMENOU, Elsie L. B. HOUNKPATIN et Brice SINSIN., 2015. Changements climatiques et gestion des ressources pastorales en zone agropastorale au Nord-Bénin : cas de la commune de Banikoara. Afrique Science, 11 (4) 201–215. Impact Factor; 3806.
- +229 95306612