Volker Steuber
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Volker Steuber has a total of 1000 co-authors. Here are the top co-authors by number of shared publications:
Neil Davey
Affiliation UH Biocomputation Research Group, Centre for Computer Science and Informatics Research, University of Hertfordshire, Hatfield United Kingdom
42 shared publicationsRod Adams
Centre for Computer Science and Informatics Research, University of Hertfordshire, College Lane, AL10 9AB Hatfield, United Kingdom
25 shared publications- Reinoud Maex
École Normale Supérieure
16 shared publications - Borys Wróbel
Laboratory of Bioengineering, Institute of Molecular Biology and Biotechnology, Adam Mickiewicz University in Poznań, Poland
10 shared publications - Maria J. Schilstra
University of Hertfordshire
10 shared publications - Christoph Metzner
Neural Information Processing Group, Institute of Software Engineering and Theoretical Computer Science, Technische Universität Berlin, Berlin, Germany
8 shared publications
Latest Publications
- Open Access
- 14 March 2026
- Published by Elsevier in Applied Soft Computing
- Vol. 197
A deep learning feature mapping algorithm for emotion detection via facial and audio signals
- 14 March 2026
- Published by Elsevier in Applied Soft Computing
- Vol. 197
Open AccessA deep learning feature mapping algorithm for emotion detection via facial and audio signals
- Open Access
- 2 February 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (2) , e0339918
Autapses enable temporal pattern recognition in spiking neural networks
Most sensory stimuli are temporal in structure. How action potentials encode the information incoming from sensory stimuli remains one of the central research questions in neuroscience. Precise spike timing is known to represent information in spiking neuronal networks, yet the information processing mechanisms of spiking neuronal networks is poorly understood. One feasible way to understand the processing mechanism of a spiking network is to associate the structural connectivity of the network with the corresponding functional behaviour. This work demonstrates the structure-function mapping of spiking networks evolved (or handcrafted) for a temporal pattern recognition task. The task is to recognise a specific order of the input signals so that the output neuron of the network spikes only for the correct placement and remains silent for all others. The minimal networks obtained for this task revealed two complementary roles of autapses in recognition. First, autapses enable a seamless transition to the next network state when a new input signal arrives. Second, in the absence of the input signal, they allow the network to maintain a network state for an extended period, a form of memory. To show that the recognition task is accomplished by transitions between network states, we map the network states of a functional spiking neural network (SNN) onto the states of a finite-state transducer (FST). Finally, based on our understanding, we define rules for constructing the topology of a network handcrafted for recognising a subsequence of signals in a particular order. The analysis of minimal networks recognising patterns of different lengths revealed a positive correlation between the pattern length and the number of autaptic connections in the network. Furthermore, in agreement with the behaviour of neurons in the network, we were able to associate specific functional roles of ’locking,’ ’switching,’ and ’accepting’ to neurons.
- 2 February 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (2) , e0339918
Open AccessAutapses enable temporal pattern recognition in spiking neural networks
Most sensory stimuli are temporal in structure. How action potentials encode the information incoming from sensory stimuli remains one of the central research questions in neuroscience. Precise spike timing is known to represent information in spiking neuronal networks, yet the information processing mechanisms of spiking neuronal networks is poorly understood. One feasible way to understand the processing mechanism of a spiking network is to associate the structural connectivity of the network with the corresponding functional behaviour. This work demonstrates the structure-function mapping of spiking networks evolved (or handcrafted) for a temporal pattern recognition task. The task is to recognise a specific order of the input signals so that the output neuron of the network spikes only for the correct placement and remains silent for all others. The minimal networks obtained for this task revealed two complementary roles of autapses in recognition. First, autapses enable a seamless transition to the next network state when a new input signal arrives. Second, in the absence of the input signal, they allow the network to maintain a network state for an extended period, a form of memory. To show that the recognition task is accomplished by transitions between network states, we map the network states of a functional spiking neural network (SNN) onto the states of a finite-state transducer (FST). Finally, based on our understanding, we define rules for constructing the topology of a network handcrafted for recognising a subsequence of signals in a particular order. The analysis of minimal networks recognising patterns of different lengths revealed a positive correlation between the pattern length and the number of autaptic connections in the network. Furthermore, in agreement with the behaviour of neurons in the network, we were able to associate specific functional roles of ’locking,’ ’switching,’ and ’accepting’ to neurons.
- Open Access
- 7 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
ConfDENSE: A conformer aware electron density based machine learning paradigm for navigating the odorant landscape
Olfaction arises from the interaction of odorants with olfactory receptors, a process shaped by molecular geometry, electron distribution, and conformational preference. We present ConfDENSE, a Set2Set enhanced PointNet model that learns directly from Hirshfeld promolecule electron-density point clouds, preserving full 3D electronic in- formation without downsampling.Despite using no receptor structural data, ConfDENSE accurately identifies bioac- tive conformers from ensemble inputs. For the only available human odorant receptor structures, the model’s selected conformers achieve sub-angstrom RMSDs to crystallo- graphic ligand poses and frequently outperform conventional docking. Combining ConfDENSE with explainability analysis further reveals the substruc- tures most responsible for receptor engagement, aligning with experimental interaction patterns. This ligand-centric and interpretable framework naturally supports phar- macophore extraction and scaffold-based design, enabling identification of conserved binding motifs even when receptor structures are missing. ConfDENSE thus provides a compact, physics-aware approach to computational olfaction, linking electron density, conformational preference, and odorant recognition in a structurally agnostic manner.
- 7 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessConfDENSE: A conformer aware electron density based machine learning paradigm for navigating the odorant landscape
Olfaction arises from the interaction of odorants with olfactory receptors, a process shaped by molecular geometry, electron distribution, and conformational preference. We present ConfDENSE, a Set2Set enhanced PointNet model that learns directly from Hirshfeld promolecule electron-density point clouds, preserving full 3D electronic in- formation without downsampling.Despite using no receptor structural data, ConfDENSE accurately identifies bioac- tive conformers from ensemble inputs. For the only available human odorant receptor structures, the model’s selected conformers achieve sub-angstrom RMSDs to crystallo- graphic ligand poses and frequently outperform conventional docking. Combining ConfDENSE with explainability analysis further reveals the substruc- tures most responsible for receptor engagement, aligning with experimental interaction patterns. This ligand-centric and interpretable framework naturally supports phar- macophore extraction and scaffold-based design, enabling identification of conserved binding motifs even when receptor structures are missing. ConfDENSE thus provides a compact, physics-aware approach to computational olfaction, linking electron density, conformational preference, and odorant recognition in a structurally agnostic manner.
- Open Access
- 5 November 2025
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 4 (11) , 3339-3350
Dense Sense: a novel approach utilizing electron density augmented machine learning paradigm to understand the complex odour landscape
Olfaction is a complex process, to navigate the complex olfactory landscape we have developed QM augmented machine learning model.
- 5 November 2025
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 4 (11) , 3339-3350
Open AccessDense Sense: a novel approach utilizing electron density augmented machine learning paradigm to understand the complex odour landscape
Olfaction is a complex process, to navigate the complex olfactory landscape we have developed QM augmented machine learning model.
- Open Access
MesoSCOUT: A novel tool for revealing mesoscale organization in the white-matter connectome
We present MesoSCOUT (Mesoscale Structural Connectome Order-complex Unveiling Topology), a framework for characterizing white matter connectivity using a method developed from computational algebraic topology: persistent homology (PH) via clique topology. Applying this method to schizophrenia, we uncover a novel, mesoscale perspective of the differences in the white-matter connectome between healthy controls (HC) and subjects with schizophrenia (SCH). We extract and compare topological motifs found in the structural connectomes of the subjects in the two groups and find significant differences. We explore the overlap of mesoscale structures found in two different datasets, COBRE (Center of Biomedical Research Excellence) and HCP(Human Connectome Project). Differences in acquisition usually render experiments recorded on different scanners incomparable, but MesoSCOUT enables cross-dataset comparisons. Our method offers a way to establish connectomic fingerprinting that could lead to a neuroimaging-based diagnosis of schizophrenia and other psychiatric and neurological conditions as well as the development of new treatments.
Open AccessMesoSCOUT: A novel tool for revealing mesoscale organization in the white-matter connectome
We present MesoSCOUT (Mesoscale Structural Connectome Order-complex Unveiling Topology), a framework for characterizing white matter connectivity using a method developed from computational algebraic topology: persistent homology (PH) via clique topology. Applying this method to schizophrenia, we uncover a novel, mesoscale perspective of the differences in the white-matter connectome between healthy controls (HC) and subjects with schizophrenia (SCH). We extract and compare topological motifs found in the structural connectomes of the subjects in the two groups and find significant differences. We explore the overlap of mesoscale structures found in two different datasets, COBRE (Center of Biomedical Research Excellence) and HCP(Human Connectome Project). Differences in acquisition usually render experiments recorded on different scanners incomparable, but MesoSCOUT enables cross-dataset comparisons. Our method offers a way to establish connectomic fingerprinting that could lead to a neuroimaging-based diagnosis of schizophrenia and other psychiatric and neurological conditions as well as the development of new treatments.
- Open Access
- 26 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
DENSE SENSE : A novel approach utilizing an electron density augmented machine learning paradigm to understand a complex odour landscape
Olfaction is a complex process which involves interaction of multiple odour receptors in nasal epithelium to produce the sensation of smell for particular odorant molecules. Elucidating structure-activity-relationships for odorants and their receptors remains difficult since crystallization of the odour receptors extremely difficult. Therefore, ligand-based approaches that leverage machine learning remain the state of the art for predicting odorant properties for molecules, such as the graph neural network approach used by. In this paper we explore how information from Quantum Mechanics (QM) could synergistically improve the results obtained with the graph neural network. Our findings underscore the possibility of this methodology in predicting odour perception directly from QM data, offering a novel approach in the Machine learning space to understand olfaction.
- 26 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessDENSE SENSE : A novel approach utilizing an electron density augmented machine learning paradigm to understand a complex odour landscape
Olfaction is a complex process which involves interaction of multiple odour receptors in nasal epithelium to produce the sensation of smell for particular odorant molecules. Elucidating structure-activity-relationships for odorants and their receptors remains difficult since crystallization of the odour receptors extremely difficult. Therefore, ligand-based approaches that leverage machine learning remain the state of the art for predicting odorant properties for molecules, such as the graph neural network approach used by. In this paper we explore how information from Quantum Mechanics (QM) could synergistically improve the results obtained with the graph neural network. Our findings underscore the possibility of this methodology in predicting odour perception directly from QM data, offering a novel approach in the Machine learning space to understand olfaction.
- Open Access
Decoding the Amplitude and Slope of Continuous Signals into Spikes with a Spiking Point Neuron Model
Scalable and efficient neural models with biological plausibility are crucial for real-time applications in neuromorphic hardware and robotics. While biophysically detailed compartmental models are valuable for their accuracy, their complexity and computational cost limit real-world application on currently available hardware. In this study, we harness the signal processing potential of the Izhikevich point neuron model to decode slope and amplitude, both important dynamical features of sensory signals. Results demonstrate that our slope-detector effectively signals up-stokes of naturalistic input signals across a wide range of frequencies. It exhibits bidirectional slope detection, with burst duration encoding slope magnitude in a graded manner. We compare to a biophysically detailed two-compartment pyramidal neuron model, showing that our bursting slope-detector performs similarly. We then demonstrate that our slope-detector does not need to burst, improving efficiency and producing more precise, discrete output by signalling events with single spikes. This makes it well-suited for real-time robotics control systems or neuromorphic hardware applications, offering greater efficiency and enabling large-scale simulations using the same computational power.
Open AccessDecoding the Amplitude and Slope of Continuous Signals into Spikes with a Spiking Point Neuron Model
Scalable and efficient neural models with biological plausibility are crucial for real-time applications in neuromorphic hardware and robotics. While biophysically detailed compartmental models are valuable for their accuracy, their complexity and computational cost limit real-world application on currently available hardware. In this study, we harness the signal processing potential of the Izhikevich point neuron model to decode slope and amplitude, both important dynamical features of sensory signals. Results demonstrate that our slope-detector effectively signals up-stokes of naturalistic input signals across a wide range of frequencies. It exhibits bidirectional slope detection, with burst duration encoding slope magnitude in a graded manner. We compare to a biophysically detailed two-compartment pyramidal neuron model, showing that our bursting slope-detector performs similarly. We then demonstrate that our slope-detector does not need to burst, improving efficiency and producing more precise, discrete output by signalling events with single spikes. This makes it well-suited for real-time robotics control systems or neuromorphic hardware applications, offering greater efficiency and enabling large-scale simulations using the same computational power.
- Open Access
Autapses enable temporal pattern recognition in spiking neural networks
Most sensory stimuli are temporal in structure. How action potentials encode the information incoming from sensory stimuli remains one of the central research questions in neuroscience. Although there is evidence that the precise timing of spikes represents information in spiking neuronal networks, information processing in spiking networks is still not fully understood. One feasible way to understand the working mechanism of a spiking network is to associate the structural connectivity of the network with the corresponding functional behaviour. This work demonstrates the structure-function mapping of spiking networks evolved (or handcrafted) for a temporal pattern recognition task. The task is to recognise a specific order of the input signals so that the Out put neurone of the network spikes only for the correct placement and remains silent for all others. The minimal networks obtained for this task revealed the twofold importance of autapses in recognition; first, autapses simplify the switching among different network states. Second, autapses enable a network to maintain a network state, a form of memory. To show that the recognition task is accomplished by transitions between network states, we map the network states of a functional spiking neural network (SNN) onto the states of a finite-state transducer (FST, a formal model of computation that generates output symbols, here: spikes or no spikes at specific times, in response to input, here: a series of input signals). Finally, based on our understanding, we define rules for constructing the topology of a network handcrafted for recognising a subsequence of signals (pattern) in a particular order. The analysis of minimal networks recognising patterns of different lengths (two to six) revealed a positive correlation between the pattern length and the number of autaptic connections in the network. Furthermore, in agreement with the behaviour of neurones in the network, we were able to associate specific functional roles of ‘locking,’ ‘switching,’ and ‘accepting’ to neurones.
Open AccessAutapses enable temporal pattern recognition in spiking neural networks
Most sensory stimuli are temporal in structure. How action potentials encode the information incoming from sensory stimuli remains one of the central research questions in neuroscience. Although there is evidence that the precise timing of spikes represents information in spiking neuronal networks, information processing in spiking networks is still not fully understood. One feasible way to understand the working mechanism of a spiking network is to associate the structural connectivity of the network with the corresponding functional behaviour. This work demonstrates the structure-function mapping of spiking networks evolved (or handcrafted) for a temporal pattern recognition task. The task is to recognise a specific order of the input signals so that the Out put neurone of the network spikes only for the correct placement and remains silent for all others. The minimal networks obtained for this task revealed the twofold importance of autapses in recognition; first, autapses simplify the switching among different network states. Second, autapses enable a network to maintain a network state, a form of memory. To show that the recognition task is accomplished by transitions between network states, we map the network states of a functional spiking neural network (SNN) onto the states of a finite-state transducer (FST, a formal model of computation that generates output symbols, here: spikes or no spikes at specific times, in response to input, here: a series of input signals). Finally, based on our understanding, we define rules for constructing the topology of a network handcrafted for recognising a subsequence of signals (pattern) in a particular order. The analysis of minimal networks recognising patterns of different lengths (two to six) revealed a positive correlation between the pattern length and the number of autaptic connections in the network. Furthermore, in agreement with the behaviour of neurones in the network, we were able to associate specific functional roles of ‘locking,’ ‘switching,’ and ‘accepting’ to neurones.
- Open Access
- 28 April 2022
- Published by Springer Nature in Schizophrenia
- Vol. 8 (1) , 1-10
The effect of alterations of schizophrenia-associated genes on gamma band oscillations
Abnormalities in the synchronized oscillatory activity of neurons in general and, specifically in the gamma band, might play a crucial role in the pathophysiology of schizophrenia. While these changes in oscillatory activity have traditionally been linked to alterations at the synaptic level, we demonstrate here, using computational modeling, that common genetic variants of ion channels can contribute strongly to this effect. Our model of primary auditory cortex highlights multiple schizophrenia-associated genetic variants that reduce gamma power in an auditory steady-state response task. Furthermore, we show that combinations of several of these schizophrenia-associated variants can produce similar effects as the more traditionally considered synaptic changes. Overall, our study provides a mechanistic link between schizophrenia-associated common genetic variants, as identified by genome-wide association studies, and one of the most robust neurophysiological endophenotypes of schizophrenia.
- 28 April 2022
- Published by Springer Nature in Schizophrenia
- Vol. 8 (1) , 1-10
Open AccessThe effect of alterations of schizophrenia-associated genes on gamma band oscillations
Abnormalities in the synchronized oscillatory activity of neurons in general and, specifically in the gamma band, might play a crucial role in the pathophysiology of schizophrenia. While these changes in oscillatory activity have traditionally been linked to alterations at the synaptic level, we demonstrate here, using computational modeling, that common genetic variants of ion channels can contribute strongly to this effect. Our model of primary auditory cortex highlights multiple schizophrenia-associated genetic variants that reduce gamma power in an auditory steady-state response task. Furthermore, we show that combinations of several of these schizophrenia-associated variants can produce similar effects as the more traditionally considered synaptic changes. Overall, our study provides a mechanistic link between schizophrenia-associated common genetic variants, as identified by genome-wide association studies, and one of the most robust neurophysiological endophenotypes of schizophrenia.
- Open Access
- 1 June 2021
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 17 (6) , e1008996
Growth rules for the repair of Asynchronous Irregular neuronal networks after peripheral lesions
Several homeostatic mechanisms enable the brain to maintain desired levels of neuronal activity. One of these, homeostatic structural plasticity, has been reported to restore activity in networks disrupted by peripheral lesions by altering their neuronal connectivity. While multiple lesion experiments have studied the changes in neurite morphology that underlie modifications of synapses in these networks, the underlying mechanisms that drive these changes are yet to be explained. Evidence suggests that neuronal activity modulates neurite morphology and may stimulate neurites to selective sprout or retract to restore network activity levels. We developed a new spiking network model of peripheral lesioning and accurately reproduced the characteristics of network repair after deafferentation that are reported in experiments to study the activity dependent growth regimes of neurites. To ensure that our simulations closely resemble the behaviour of networks in the brain, we model deafferentation in a biologically realistic balanced network model that exhibits low frequency Asynchronous Irregular (AI) activity as observed in cerebral cortex. Our simulation results indicate that the re-establishment of activity in neurons both within and outside the deprived region, the Lesion Projection Zone (LPZ), requires opposite activity dependent growth rules for excitatory and inhibitory post-synaptic elements. Analysis of these growth regimes indicates that they also contribute to the maintenance of activity levels in individual neurons. Furthermore, in our model, the directional formation of synapses that is observed in experiments requires that pre-synaptic excitatory and inhibitory elements also follow opposite growth rules. Lastly, we observe that our proposed structural plasticity growth rules and the inhibitory synaptic plasticity mechanism that also balances our AI network both contribute to the restoration of the network to pre-deafferentation stable activity levels.
- 1 June 2021
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 17 (6) , e1008996
Open AccessGrowth rules for the repair of Asynchronous Irregular neuronal networks after peripheral lesions
Several homeostatic mechanisms enable the brain to maintain desired levels of neuronal activity. One of these, homeostatic structural plasticity, has been reported to restore activity in networks disrupted by peripheral lesions by altering their neuronal connectivity. While multiple lesion experiments have studied the changes in neurite morphology that underlie modifications of synapses in these networks, the underlying mechanisms that drive these changes are yet to be explained. Evidence suggests that neuronal activity modulates neurite morphology and may stimulate neurites to selective sprout or retract to restore network activity levels. We developed a new spiking network model of peripheral lesioning and accurately reproduced the characteristics of network repair after deafferentation that are reported in experiments to study the activity dependent growth regimes of neurites. To ensure that our simulations closely resemble the behaviour of networks in the brain, we model deafferentation in a biologically realistic balanced network model that exhibits low frequency Asynchronous Irregular (AI) activity as observed in cerebral cortex. Our simulation results indicate that the re-establishment of activity in neurons both within and outside the deprived region, the Lesion Projection Zone (LPZ), requires opposite activity dependent growth rules for excitatory and inhibitory post-synaptic elements. Analysis of these growth regimes indicates that they also contribute to the maintenance of activity levels in individual neurons. Furthermore, in our model, the directional formation of synapses that is observed in experiments requires that pre-synaptic excitatory and inhibitory elements also follow opposite growth rules. Lastly, we observe that our proposed structural plasticity growth rules and the inhibitory synaptic plasticity mechanism that also balances our AI network both contribute to the restoration of the network to pre-deafferentation stable activity levels.