Contextual signal processing in the visual system
Neurons in the
early visual areas of higher animals respond selectively to local
properties of images as contrast, movement, texture and so on. These
neurons are part of bigger ensembles (cortical columns, layers or
areas) which represent special aspects of visual information processing
and are defined under physiological or anatomical point of view.
In the project B2 we investigate models for contextual signal processing in the visual system of higher animals. Recently, the spatial and temporal response properties of cortical neurons in the visual system are experimentally well documented. We develop computational models which are based on these experimental findings and test their ability to generate the observed physiological properties as for example orientation selectivity.
On the one hand we use effective models of neurons for numerical simulations to predict the outcome of experiments. We derive these effective models of cortical neurons from detailed modeling studies of "conductance based" type of neurons. On the other hand we use these models to investigate the hypothesis of optimal information coding and we explore the applicability of optimality criteria like Infomax. Therefore we develop statistical descriptions of natural visual stimuli and use them to calibrate our models.
Furthermore, we cooperate with departments of part A to prove and adjust our "Data Mining" algorithms for micro array data analysis.
In the project B2 we investigate models for contextual signal processing in the visual system of higher animals. Recently, the spatial and temporal response properties of cortical neurons in the visual system are experimentally well documented. We develop computational models which are based on these experimental findings and test their ability to generate the observed physiological properties as for example orientation selectivity.
On the one hand we use effective models of neurons for numerical simulations to predict the outcome of experiments. We derive these effective models of cortical neurons from detailed modeling studies of "conductance based" type of neurons. On the other hand we use these models to investigate the hypothesis of optimal information coding and we explore the applicability of optimality criteria like Infomax. Therefore we develop statistical descriptions of natural visual stimuli and use them to calibrate our models.
Furthermore, we cooperate with departments of part A to prove and adjust our "Data Mining" algorithms for micro array data analysis.
german version |