Spatial Attention Improves Object Localization: A Biologically Plausible Neuro-Computational Model for Use in Virtual Reality

Jamalian, Amirhossein; Bergelt, Julia; Dinkelbach, Helge Ülo

doi:10.1109/iccvw.2017.320

Cited by 3 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For rate-coded models, this could be formats which use structural properties, such as the diagonal format. Some neuro-computational models developed in our lab (e.g., Jamalian et al, 2017 ) contain matrices which have a banded matrix structure. A promising direction may be the implementation of sliced matrix formats (e.g., Kreutzer et al, 2014 ).…”

Section: Discussionmentioning

confidence: 99%

Auto-Selection of an Optimal Sparse Matrix Format in the Neuro-Simulator ANNarchy

Dinkelbach

Bouhlal

Vitay

et al. 2022

Front. Neuroinform.

Self Cite

View full text Add to dashboard Cite

Modern neuro-simulators provide efficient implementations of simulation kernels on various parallel hardware (multi-core CPUs, distributed CPUs, GPUs), thereby supporting the simulation of increasingly large and complex biologically realistic networks. However, the optimal configuration of the parallel hardware and computational kernels depends on the exact structure of the network to be simulated. For example, the computation time of rate-coded neural networks is generally limited by the available memory bandwidth, and consequently, the organization of the data in memory will strongly influence the performance for different connectivity matrices. We pinpoint the role of sparse matrix formats implemented in the neuro-simulator ANNarchy with respect to computation time. Rather than asking the user to identify the best data structures required for a given network and platform, such a decision could also be carried out by the neuro-simulator. However, it requires heuristics that need to be adapted over time for the available hardware. The present study investigates how machine learning methods can be used to identify appropriate implementations for a specific network. We employ an artificial neural network to develop a predictive model to help the developer select the optimal sparse matrix format. The model is first trained offline using a set of training examples on a particular hardware platform. The learned model can then predict the execution time of different matrix formats and decide on the best option for a specific network. Our experimental results show that using up to 3,000 examples of random network configurations (i.e., different population sizes as well as variable connectivity), our approach effectively selects the appropriate configuration, providing over 93% accuracy in predicting the suitable format on three different NVIDIA devices.

show abstract

Section: Discussionmentioning

confidence: 99%

Auto-Selection of an Optimal Sparse Matrix Format in the Neuro-Simulator ANNarchy

Dinkelbach

Bouhlal

Vitay

et al. 2022

Front. Neuroinform.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Network models in ANNarchy are defined through equations written in "natural language". ANNarchy has been used to implement models of the BG pathways (Baladron et al, 2019;Gönner et al, 2020;Maith et al, 2020;Villagrasa et al, 2018), spatial attention and vision (Bergelt & Hamker, 2019;Jamalian et al, 2017;Larisch et al, 2020) and learning and memory (Gönner et al, 2017;J. Müller et al, 2018;Schmid et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

Virtual deep brain stimulation: Multiscale co-simulation of a spiking basal ganglia model and a whole-brain mean-field model with The Virtual Brain

Meier

Perdikis

Blickensdörfer

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Deep brain stimulation (DBS) has been successfully applied in various neurodegenerative diseases as an effective symptomatic treatment. However, its mechanisms of action within the brain network are still poorly understood. Many virtual DBS models analyze a subnetwork around the basal ganglia and its dynamics as a spiking network with their details validated by experimental data. However, connectomic evidence shows widespread effects of DBS affecting many different cortical and subcortical areas. From a clinical perspective, various effects of DBS besides the motoric impact have been demonstrated. The neuroinformatics platform The Virtual Brain (TVB) offers a modeling framework allowing us to virtually perform stimulation, including DBS, and forecast the outcome from a dynamic systems perspective prior to invasive surgery with DBS lead placement. For an accurate prediction of the effects of DBS, we implement a detailed spiking model of the basal ganglia, which we combine with TVB via our previously developed co-simulation environment. This multiscale co-simulation approach builds on the extensive previous literature of spiking models of the basal ganglia while simultaneously offering a whole-brain perspective on widespread effects of the stimulation going beyond the motor circuit. In the first demonstration of our model, we show that virtual DBS can move the firing rates of a Parkinson’s disease patient’s thalamus - basal ganglia network towards the healthy regime while, at the same time, altering the activity in distributed cortical regions with a pronounced effect in frontal regions. Thus, we provide proof of concept for virtual DBS in a co-simulation environment with TVB. The developed modeling approach has the potential to optimize DBS lead placement and configuration and forecast the success of DBS treatment for individual patients.Highlights-We implement and validate a co-simulation approach of a spiking network model for subcortical regions in and around the basal ganglia and interface it with mean-field network models for each cortical region.-Our simulations are based on a normative connectome including detailed tracts between the cortex and the basal ganglia regions combined with subject-specific optimized weights for a healthy control and a patient with Parkinson’s disease.-We provide proof of concept by demonstrating that the implemented model shows biologically plausible dynamics during resting state including decreased thalamic activity in the virtual patient and during virtual deep brain stimulation including normalized thalamic activity and distributed altered cortical activity predominantly in frontal regions.-The presented co-simulation model can be used to tailor deep brain stimulation for individual patients.

show abstract

Multi-modal Spatial Object Localization in Virtual Reality for Deaf and Hard-of-Hearing People

Mirzaei

Kán

Kaufmann

2021

2021 IEEE Virtual Reality and 3D User Interfaces (VR)

View full text Add to dashboard Cite

Spatial Attention Improves Object Localization: A Biologically Plausible Neuro-Computational Model for Use in Virtual Reality

Cited by 3 publications

References 18 publications

Auto-Selection of an Optimal Sparse Matrix Format in the Neuro-Simulator ANNarchy

Auto-Selection of an Optimal Sparse Matrix Format in the Neuro-Simulator ANNarchy

Virtual deep brain stimulation: Multiscale co-simulation of a spiking basal ganglia model and a whole-brain mean-field model with The Virtual Brain

Multi-modal Spatial Object Localization in Virtual Reality for Deaf and Hard-of-Hearing People

Contact Info

Product

Resources

About