Cristian Axenie's repositories
corr-learn-som-quadrotor
Quadrotor implementation of the distributed cognitive maps for environment interpretation. Use artificial data selected / given by user. The input activity mildly influences the belief of the cortically inspired relaxation network. The network dynamics ensures convergence to a stable state in which all quantities are in agreement given the mathematical constraints. The constraints are learned through biologically plausible mechanisms from data. Sensory correlation learning neural network. Unsupervised learning of functional relationships between 2 input sensory streams. The network uses SOMs to encode the input data into a population code and Hebbian learning to extract the co-activation patterns encoding the function. Datasets from quadrotor system.
gas-antenna-demo
Sample demo software for Genetic Algorithms for RF antenna design. The task is to find an optimal wire antenna design (shape) knowing the number of points, the frequency and the desired gain.
spartan6-linux-dev
Spartan6-LX9 MICROBoard Getting started with Linux Embedded.
gas-opt-demo
Simple Genetic Algorithm for function optimization demo.
unsupervised-relation-learning
Unsupervised learning of relations following the network combining WTA, HL and HAR over the projections sharpening architecture. Designed by Cook et al.
artemic
ARTEMIC = Advanced Real-time Embedded Mobile robot Intelligent ControllerReal-time Linux fault tolerant control application for mobile robots. Multi-level control application: Fuzzy Sliding-Mode kinematic controller, EKF fault tolerance module and ultrasound based SLAM module.
code-snippets
Various code snippets
corr-learn-som
Simple implementation of the distributed cognitive maps for environment interpretation.
dev-sensor-fusion-net
Unsupervised sensory correlation learning using Self-Organizing-Maps for sensor fusion. Each sensory variable projects onto a SOM network which depending on the global network connectivity (1, 2, ... , N vars) connects to other SOM associated with other sensory variables. Extension to extract also temporal correlations using a recursive architecture.
embedded-ser2eth-converter
Simple embededed serial (USART) to Ethernet (TCP/IP) converter using AVR32 and ATMega16 platforms. Development scheme for industrial automation and supervisory control and distributed data acquisition systems.
hopfield-distort-demo
A very simple example of a recurrent Hopfield network applied to image undistortion.
neural-net-anatomy
Neural Network Anatomy Study
oom-oop-intro
Einführung in die objektorientierte Modellierung und Programmierung an einem Beispiel aus der Produktion
corr-learn-som-python-demo
About Unsupervised learning of relations following the network combining WTA, HL and HAR over the projections sharpening architecture. Python jupyter demo.
cristianaxenie.github.io
Github Pages template for academic personal websites, forked from mmistakes/minimal-mistakes
developing-fusion-network-fast
Simple implementation of the distributed cognitive maps for environment interpretation. The input activity mildly influences the belief of the cortically inspired relaxation network. The network dynamics ensures convergence to a stable state in which all quantities are in agreement given the mathematical constraints. The constraints are learned through biologically plausible mechanisms from data. Sensory correlation learning neural network. Unsupervised learning of functional relationships between 2 input sensory streams. The network uses SOMs to encode the input data into a population code and Hebbian learning to extract the co-activation patterns encoding the function. Modelling human circuitry development for multisensory integration. C code implementation.
dynamic-association-net
Dynamic association model to learn correlations between paired modalities samples acquired from different sensors current scenario is using only 2 input modalities.
fusion-maps-population-code
Demo software usign population code for estimating arbitrary functions. The setup contains 2 input populations each coding some scalar (unimodal) variable which are projected onto a 2D network of units with neurons exhibiting short range excitation and long range inhibition dynamics. The ouput from the intermediate layer is projected to an output population the network has no explicit input and output as each of the populations may be considered inputs / outputs and the processing happens in the intermediate layer.
hopfield-memory-demo
Simple auto-associative memory to recover names and phone numbers and/or match them using a Hopfield network.
ibfn-net-cue-integration
Population code based network for cue integration using Basis Function Neurons. Simple case scenario for 2 inputs.
ip-camera-overhead-tracker
Overhead ip cam tracker (single cam) using OpenCV acquisition and processing of MJPEG/RTSP streams.
learning-multisensory-prediction
Learning multisensory prediction for road traffic
msp430-android-gyro-acc-tester
Simple embedded platform to test inertial sensors (e.g. Android phones) with TI MSP430 MCU.
neuro-drone-daq-ctrl
The software infrastructure for data acquisition, synchronization with 3D tracker system and logging was developed at TUM.
neurorobotics.me_backend
Backend and admin for neurorobotics.me
predictive-maintenance-learner
Neural Network Predictive Maintenance System
sensory-projections-som
Pointwise organized projections of sensory afferents. Network to simulate the organization of pointwise connections from sensory afferents in cortex. Work of Kohonen, 2005/2006: Self-organizing neural projections. Pointwise organizing projections.
sharp-learning
Simple implementation of the sharp learning algorithm using the connectivity presented in Cook et. al.
sigma-pi-som-learning
Implementation of a network which learns coordinate systems transformations using sigma-pi units that self-organize (Weber et al.).
spices-lab-puzzlebot-dev
This repository contains all the files required for the Puzzlebot Development