promanev
I am working as a Data Scientist at WEX Inc tackling the prediction of marketing campaign success, customer attrition, and fraud prevention.
WEX Inc.Portland, ME
Pinned Repositories
Beta_ERD_SNN_GA
My thesis project that focuses on a computational, embodied model of bipedal robots with neural controllers based on spiking neurons. The goal of the project is to test the function of beta event-related desynchronization, a phenomenon exhibited by the human motor system with yet unclear function.
Forex-Price-Prediction-with-LSTM
Initial commit
PDSTEP
Code for PDSTEP project (Winter 2015-2016) that used CTRNN-based controller operating a simulated bipedal robot (C++) in Bullet physics engine, DE genetic algorithm (MATLAB)
PDSTEP-Matlab-code
Matlab code that runs evolution for PDSTEP project (Winter 2015-2016 version). Contains two different evolutionary algorithms (DE and CMAES) and a number of scripts that were used to export and analyze kinematic data
PDSTEP-SNN-version
The most up-to-date version of PDSTEP project (Summer 2016 - ongoing). Project is now focused on testing a hypothesis about the functional meaning of a phenomenon observed in the human motion system - beta-band oscillations and beta event-related desynchronization (beta ERD) - which is used as an indicator of cortical preparation for a change in motor states. There is an on-going debate about the actual meaning of beta ERD and I am hoping to shed some light on it. The neural model is based on a spiking neural network with Izhikevich model neurons (Izhikevich, 2003) operating a bipedal robot in Bullet physics engine. The model is optimized through a search for a synaptic connection matrix that achieves target behavior. The search is guided by a genetic algorithm evolving bit strings similar to the work of Dario Floreano (e.g., Floreano et al., 2005).
PDSTEP_SNN_PEAS
HyperNEAT-aided PDSTEP with SNNs
promanev's Repositories
promanev/PDSTEP-SNN-version
The most up-to-date version of PDSTEP project (Summer 2016 - ongoing). Project is now focused on testing a hypothesis about the functional meaning of a phenomenon observed in the human motion system - beta-band oscillations and beta event-related desynchronization (beta ERD) - which is used as an indicator of cortical preparation for a change in motor states. There is an on-going debate about the actual meaning of beta ERD and I am hoping to shed some light on it. The neural model is based on a spiking neural network with Izhikevich model neurons (Izhikevich, 2003) operating a bipedal robot in Bullet physics engine. The model is optimized through a search for a synaptic connection matrix that achieves target behavior. The search is guided by a genetic algorithm evolving bit strings similar to the work of Dario Floreano (e.g., Floreano et al., 2005).
promanev/Forex-Price-Prediction-with-LSTM
Initial commit
promanev/PDSTEP
Code for PDSTEP project (Winter 2015-2016) that used CTRNN-based controller operating a simulated bipedal robot (C++) in Bullet physics engine, DE genetic algorithm (MATLAB)
promanev/Beta_ERD_SNN_GA
My thesis project that focuses on a computational, embodied model of bipedal robots with neural controllers based on spiking neurons. The goal of the project is to test the function of beta event-related desynchronization, a phenomenon exhibited by the human motor system with yet unclear function.
promanev/PDSTEP-Matlab-code
Matlab code that runs evolution for PDSTEP project (Winter 2015-2016 version). Contains two different evolutionary algorithms (DE and CMAES) and a number of scripts that were used to export and analyze kinematic data
promanev/PDSTEP_SNN_PEAS
HyperNEAT-aided PDSTEP with SNNs