Johan Reimann Email & Phone Number

Niskayuna, Ny

Research and Development of Deep Learning, Reinforcement Learning and Inverse Reinforcement Learning methods and solutions.

Research and development of scalable machine learning algorithms for asset management, maintenance planning and data exploration.

Niskayuna, Ny

Research and development of scalable machine learning algorithms for asset management, maintenance planning and data exploration.Algorithm Development Efforts:Deep Learning Technologies based on Restricted Boltzmann Machines: Developed a deep learning tool that can be used as a generative modeling tool to impute missing values. Algorithm speedup was accomplished using GPU parallelization.Implementation Language: Java, OpenCLMassively Parallel Particle Filtering Framework: mplemented a massively parallel particle filtering based forecasting framework used to propagate and forecasting component life/health given the surface of a finite element model. Implementation Language: Java, OpenCL.Minimum Spanning-tree based clustering: Using minimum spanning trees to capture data structure information and automatically determine the number of clusters in a data set. Implementation Language: Matlab..

Research and development of scheduling, classification, prognostic and diagnostic algorithms with emphasis on establishing theoretical performance estimates. Application domains: Mechanical system failures, image processing, maintenance planning and supply chain management.Algorithm Development Efforts:Symbolic Regression Tool using Genetic Programming:Developed a tool that searches for closed form mathematical expressions that explains the relationships between variables. The toolset is highly customizable and leverages the parallel processing capabilities of multi-core processing platforms. Implementation Language: Matlab.Dynamic Bayesian Belief Network Implementation: Implemented a Dynamic Bayesian Belief Network using Matlab. The framework is able to capture the joint probability distribution from time series data and perform exact inference on new measurements provided to the network.Multi-objective Particle Swarm Optimization Algorithm:Solves general constrained multi-objective problems using the particle swarm optimization technique. Algorithm was implemented in Matlab and allows for a mixture of integer and continuous objective and constraint functions.

Rochester, New York Area

Course work covers mathematical modeling of mechanical, electrical, fluid and mixed systems and advanced control system design.

Atlanta, Ga

Developed optimal control algorithms in the intelligent control laboratory in an effort to improve the autonomy of individual and swarms of unmanned aerial vehicles (UAVs).

Ph.D, Electrical And Computer Engineering

B.S, Electrical Engineering And Mathematics