Search

Your search keyword '"Scholte, H. Steven"' showing total 375 results
Start Over Author "Scholte, H. Steven"
375 results on '"Scholte, H. Steven"'

1. LocalNorm: Robust Image Classification through Dynamically Regularized Normalization

Author

Yin, Bojian, Schaafsma, Siebren, Corporaal, Henk, Scholte, H. Steven, and Bohte, Sander M.

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

While modern convolutional neural networks achieve outstanding accuracy on many image classification tasks, they are, compared to humans, much more sensitive to image degradation. Here, we describe a variant of Batch Normalization, LocalNorm, that regularizes the normalization layer in the spirit of Dropout while dynamically adapting to the local image intensity and contrast at test-time. We show that the resulting deep neural networks are much more resistant to noise-induced image degradation, improving accuracy by up to three times, while achieving the same or slightly better accuracy on non-degraded classical benchmarks. In computational terms, LocalNorm adds negligible training cost and little or no cost at inference time, and can be applied to already-trained networks in a straightforward manner., Comment: 14 pages, 17 figures

Published
2019

3. Efficient Computation in Adaptive Artificial Spiking Neural Networks

Author

Zambrano, Davide, Nusselder, Roeland, Scholte, H. Steven, and Bohte, Sander

Subjects
Computer Science - Neural and Evolutionary Computing
Abstract

Artificial Neural Networks (ANNs) are bio-inspired models of neural computation that have proven highly effective. Still, ANNs lack a natural notion of time, and neural units in ANNs exchange analog values in a frame-based manner, a computationally and energetically inefficient form of communication. This contrasts sharply with biological neurons that communicate sparingly and efficiently using binary spikes. While artificial Spiking Neural Networks (SNNs) can be constructed by replacing the units of an ANN with spiking neurons, the current performance is far from that of deep ANNs on hard benchmarks and these SNNs use much higher firing rates compared to their biological counterparts, limiting their efficiency. Here we show how spiking neurons that employ an efficient form of neural coding can be used to construct SNNs that match high-performance ANNs and exceed state-of-the-art in SNNs on important benchmarks, while requiring much lower average firing rates. For this, we use spike-time coding based on the firing rate limiting adaptation phenomenon observed in biological spiking neurons. This phenomenon can be captured in adapting spiking neuron models, for which we derive the effective transfer function. Neural units in ANNs trained with this transfer function can be substituted directly with adaptive spiking neurons, and the resulting Adaptive SNNs (AdSNNs) can carry out inference in deep neural networks using up to an order of magnitude fewer spikes compared to previous SNNs. Adaptive spike-time coding additionally allows for the dynamic control of neural coding precision: we show how a simple model of arousal in AdSNNs further halves the average required firing rate and this notion naturally extends to other forms of attention. AdSNNs thus hold promise as a novel and efficient model for neural computation that naturally fits to temporally continuous and asynchronous applications.

Published
2017

4. Visual pathways from the perspective of cost functions and multi-task deep neural networks

Author

Scholte, H. Steven, Losch, Max M., Ramakrishnan, Kandan, de Haan, Edward H. F., and Bohte, Sander M.

Subjects
Quantitative Biology - Neurons and Cognition
Abstract

Vision research has been shaped by the seminal insight that we can understand the higher-tier visual cortex from the perspective of multiple functional pathways with different goals. In this paper, we try to give a computational account of the functional organization of this system by reasoning from the perspective of multi-task deep neural networks. Machine learning has shown that tasks become easier to solve when they are decomposed into subtasks with their own cost function. We hypothesize that the visual system optimizes multiple cost functions of unrelated tasks and this causes the emergence of a ventral pathway dedicated to vision for perception, and a dorsal pathway dedicated to vision for action. To evaluate the functional organization in multi-task deep neural networks, we propose a method that measures the contribution of a unit towards each task, applying it to two networks that have been trained on either two related or two unrelated tasks, using an identical stimulus set. Results show that the network trained on the unrelated tasks shows a decreasing degree of feature representation sharing towards higher-tier layers while the network trained on related tasks uniformly shows high degree of sharing. We conjecture that the method we propose can be used to analyze the anatomical and functional organization of the visual system and beyond. We predict that the degree to which tasks are related is a good descriptor of the degree to which they share downstream cortical-units., Comment: 16 pages, 5 figures

Published
2017
Full Text
View/download PDF

5. LocalNorm: Robust Image Classification Through Dynamically Regularized Normalization

Author

Yin, Bojian, Scholte, H. Steven, Bohté, Sander, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Farkaš, Igor, editor, Masulli, Paolo, editor, Otte, Sebastian, editor, and Wermter, Stefan, editor

Published
2021
Full Text
View/download PDF

10. Exome Chip Meta-analysis Fine Maps Causal Variants and Elucidates the Genetic Architecture of Rare Coding Variants in Smoking and Alcohol Use

Author

Surendran, Praveen, Young, Robin, Barnes, Daniel R., Nielsen, Sune Fallgaard, Rasheed, Asif, Samuel, Maria, Zhao, Wei, Kontto, Jukka, Perola, Markus, Caslake, Muriel, de Craen, Anton J.M., Trompet, Stella, Uria-Nickelsen, Maria, Malarstig, Anders, Reily, Dermot F., Hoek, Maarten, Vogt, Thomas, Jukema, J. Wouter, Sattar, Naveed, Ford, Ian, Packard, Chris J., Alam, Dewan S., Majumder, Abdulla al Shafi, Di Angelantonio, Emanuele, Chowdhury, Rajiv, Amouyel, Philippe, Arveiler, Dominique, Blankenberg, Stefan, Ferrières, Jean, Kee, Frank, Kuulasmaa, Kari, Müller-Nurasyid, Martina, Veronesi, Giovanni, Virtamo, Jarmo, EPIC-CVD Consortium, Frossard, Philippe, Nordestgaard, Børge Grønne, Saleheen, Danish, Danesh, John, Butterworth, Adam S., Howson, Joanna M.M., Erzurumluoglu, A. Mesut, Jackson, Victoria E., Melbourne, Carl A., Varga, Tibor V., Warren, Helen R., Tragante, Vinicius, Tachmazidou, Ioanna, Harris, Sarah E., Evangelou, Evangelos, Marten, Jonathan, Zhang, Weihua, Altmaier, Elisabeth, Luan, Jian’an, Langenberg, Claudia, Scott, Robert A., Yaghootkar, Hanieh, Stirrups, Kathleen, Kanoni, Stavroula, Marouli, Eirini, Karpe, Fredrik, Dominiczak, Anna F., Sever, Peter, Poulter, Neil, Rolandsson, Olov, Baumbach, Clemens, Afaq, Saima, Chambers, John C., Kooner, Jaspal S., Wareham, Nicholas J., Renström, Frida, Hallmans, Göran, Marioni, Riccardo E., Corley, Janie, Starr, John M., Verweij, Niek, de Boer, Rudolf A., van der Meer, Peter, Yavas, Ersin, Vaartjes, Ilonca, Bots, Michiel L., Asselbergs, Folkert W., Grabe, Hans J., Völzke, Henry, Nauck, Matthias, Weiss, Stefan, Pharoah, Paul D.P., Dunning, Alison M., Dennis, Joe G., Thompson, Deborah J., Michailidou, Kyriaki, Easton, Douglas F., Antoniou, Antonis C., Tyrrell, Jessica, Mihailov, Evelin, Samani, Nilesh J., Zhou, Kaixin, Neville, Matthew J., Metspalu, Andres, Palmer, Colin N.A., Hall, Ian P., Strachan, David P., Deary, Ian J., Frayling, Tim M., Hayward, Caroline, van der Harst, Pim, Zeggini, Eleftheria, Understanding Society Scientific Group, Munroe, Patricia B., Jansson, Jan-Håkan, Franks, Paul W., Deloukas, Panos, Caulfield, Mark J., Wain, Louise V., Tobin, Martin D., Brazel, David M., Jiang, Yu, Hughey, Jordan M., Turcot, Valérie, Zhan, Xiaowei, Gong, Jian, Batini, Chiara, Weissenkampen, J. Dylan, Liu, MengZhen, Bertelsen, Sarah, Chou, Yi-Ling, Faul, Jessica D., Haessler, Jeff, Hammerschlag, Anke R., Hsu, Chris, Kapoor, Manav, Lai, Dongbing, Le, Nhung, de Leeuw, Christiaan A., Loukola, Anu, Mangino, Massimo, Pistis, Giorgio, Qaiser, Beenish, Rohde, Rebecca, Shao, Yaming, Stringham, Heather, Wetherill, Leah, Agrawal, Arpana, Bierut, Laura, Chen, Chu, Eaton, Charles B., Goate, Alison, Haiman, Christopher, Heath, Andrew, Iacono, William G., Martin, Nicholas G., Polderman, Tinca J., Reiner, Alex, Rice, John, Schlessinger, David, Scholte, H. Steven, Smith, Jennifer A., Tardif, Jean-Claude, Tindle, Hilary A., van der Leij, Andries R., Boehnke, Michael, Chang-Claude, Jenny, Cucca, Francesco, David, Sean P., Foroud, Tatiana, Kardia, Sharon L.R., Kooperberg, Charles, Laakso, Markku, Lettre, Guillaume, Madden, Pamela, McGue, Matt, North, Kari, Posthuma, Danielle, Spector, Timothy, Stram, Daniel, Weir, David R., Kaprio, Jaakko, Abecasis, Gonçalo R., Liu, Dajiang J., and Vrieze, Scott

Published
2019
Full Text
View/download PDF

19. Dyslexia and Voxel-Based Morphometry: Correlations between Five Behavioural Measures of Dyslexia and Gray and White Matter Volumes

Author

Tamboer, Peter, Scholte, H. Steven, and Vorst, Harrie C. M.

Abstract

In voxel-based morphometry studies of dyslexia, the relation between causal theories of dyslexia and gray matter (GM) and white matter (WM) volume alterations is still under debate. Some alterations are consistently reported, but others failed to reach significance. We investigated GM alterations in a large sample of Dutch students (37 dyslexics and 57 non-dyslexics) with two analyses: group differences in local GM and total GM and WM volume and correlations between GM and WM volumes and five behavioural measures. We found no significant group differences after corrections for multiple comparisons although total WM volume was lower in the group of dyslexics when age was partialled out. We presented an overview of uncorrected clusters of voxels (p??200) with reduced or increased GM volume. We found four significant correlations between factors of dyslexia representing various behavioural measures and the clusters found in the first analysis. In the whole sample, a factor related to performances in spelling correlated negatively with GM volume in the left posterior cerebellum. Within the group of dyslexics, a factor related to performances in Dutch-English rhyme words correlated positively with GM volume in the left and right caudate nucleus and negatively with increased total WM volume. Most of our findings were in accordance with previous reports. A relatively new finding was the involvement of the caudate nucleus. We confirmed the multiple cognitive nature of dyslexia and suggested that experience greatly influences anatomical alterations depending on various subtypes of dyslexia, especially in a student sample.

Published
2015
Full Text
View/download PDF

20. Are Autistic Traits in the General Population Related to Global and Regional Brain Differences?

Author

Koolschijn, P. Cédric M. P., Geurts, Hilde M., van der Leij, Andries R., and Scholte, H. Steven

Abstract

There is accumulating evidence that autistic-related traits in the general population lie on a continuum, with autism spectrum disorders representing the extreme end of this distribution. Here, we tested the hypothesis of a possible relationship between autistic traits and brain morphometry in the general population. Participants completed the short autism-spectrum quotient-questionnaire (AQ); T1-anatomical and DWI-scans were acquired. Associations between autistic traits and gray matter, and white matter microstructural-integrity were performed on the exploration-group (N = 204; 105 males, M-age = 22.85), and validated in the validation-group (N = 304; 155 males, M-age = 22.82). No significant associations were found between AQ-scores and brain morphometry in the exploration-group, or after pooling the data. This questions the assumption that autistic traits and their morphological associations do lie on a continuum in the general population.

Published
2015
Full Text
View/download PDF

26. The Relationship between Grey-Matter and ASD and ADHD Traits in Typical Adults

Author

Geurts, Hilde M., Ridderinkhof, K. Richard, and Scholte, H. Steven

Abstract

We tested whether in 85 healthy adults (18-29 years) there is a relationship between grey-matter (GM) volume and autism and ADHD symptom severity. The structural MRI findings and autism and ADHD self-reports revealed that autism and ADHD symptom severity was correlated with GM volume in the left inferior frontal gyrus. Autism symptom-severity was correlated with the left posterior cingulate, ADHD with the right parietal lobe, right temporal frontal cortex, bilateral thalamus, and left hippocampus/amygdala complex. Symptom severity of both disorders form a continuum extending into the general population, but it seems to be an oversimplification to typify psychiatric disorders such as autism and ADHD solely as extremes of brain structure abnormalities.

Published
2013
Full Text
View/download PDF

28. Coherent versus Component Motion Perception in Autism Spectrum Disorder

Author

Vandenbroucke, Myriam W. G., Scholte, H. Steven, and van Engeland, Herman

Abstract

Research on visual perception in Autism Spectrum Disorder (ASD) tries to reveal the underlying mechanisms of aberrant local and global processing. Global motion perception is one way to study this aspect of ASD. We used plaid motion stimuli, which can be perceived as a coherently moving pattern, requiring feature integration, or as two transparent gratings sliding over each other. If global motion detection is impaired in ASD, this would lead to a decrease of the total time that a coherent pattern is perceived. However, in contrast to other studies in the literature, our results gave no evidence of impaired global motion perception in people with ASD. A reconciliation of the different outcomes is proposed based on spatial frequency processing in ASD.

Published
2008
Full Text
View/download PDF

29. A Neural Substrate for Atypical Low-Level Visual Processing in Autism Spectrum Disorder

Author

Vandenbroucke, Myriam W. G., Scholte, H. Steven, and van Engeland, Herman

Abstract

An important characteristic of autism spectrum disorder (ASD) is increased visual detail perception. Yet, there is no standing neurobiological explanation for this aspect of the disorder. We show evidence from EEG data, from 31 control subjects (three females) and 13 subjects (two females) aged 16-28 years, for a specific impairment in object boundary detection in ASD, which is present as early as 120 ms after stimulus presentation. In line with a neural network model explicating the role of feedforward, horizontal and recurrent processing in visual perception, we can attribute this deficit to a dysfunction of horizontal connections within early visual areas. Interestingly, ASD subjects showed an increase in subsequent activity at lateral occipital sites (225 ms), which might reflect a compensational mechanism. In contrast, recurrent processing between higher and lower visual areas (around 260 ms), associated with the segregation between figure and background, was normal. Our results show specific neural abnormalities in ASD related to low-level visual processing. In addition, given the reconciliation between our findings and previous neuropathology and neurochemistry research, we suggest that atypical horizontal interactions might reflect a more general neural abnormality in this disorder.

Published
2008
Full Text
View/download PDF

33. Leveraging Spiking Deep Neural Networks to Understand the Neural Mechanisms Underlying Selective Attention

Author

Sörensen, Lynn K.A., Zambrano, Davide, Slagter, Heleen A., Bohté, Sander M., Scholte, H. Steven, Sörensen, Lynn K.A., Zambrano, Davide, Slagter, Heleen A., Bohté, Sander M., and Scholte, H. Steven

Abstract

Spatial attention enhances sensory processing of goalrelevant information and improves perceptual sensitivity. Yet, the specific neural mechanisms underlying the effects of spatial attention on performance are still contested. Here, we examine different attention mechanisms in spiking deep convolutional neural networks. We directly contrast effects of precision (internal noise suppression) and two different gain modulation mechanisms on performance on a visual search task with complex real-world images. Unlike standard artificial neurons, biological neurons have saturating activation functions, permitting implementation of attentional gain as gain on a neuron’s input or on its outgoing connection. We show that modulating the connection is most effective in selectively enhancing information processing by redistributing spiking activity and by introducing additional task-relevant information, as shown by representational similarity analyses. Precision only produced minor attentional effects in performance. Our results, which mirror empirical findings, show that it is possible to adjudicate between attention mechanisms using more biologically realistic models and natural stimuli.

Published
2022
Full Text
View/download PDF

34. A Critical Test of Deep Convolutional Neural Networks' Ability to Capture Recurrent Processing in the Brain Using Visual Masking

Author

Loke, Jessica, Seijdel, Noor, Snoek, Lukas, Meer, Matthew van der, van de Klundert, Ron, Quispel, Eva, Cappaert, Natalie, Scholte, H. Steven, Loke, Jessica, Seijdel, Noor, Snoek, Lukas, Meer, Matthew van der, van de Klundert, Ron, Quispel, Eva, Cappaert, Natalie, and Scholte, H. Steven

Abstract

Recurrent processing is a crucial feature in human visual processing supporting perceptual grouping, figure-ground segmentation, and recognition under challenging conditions. There is a clear need to incorporate recurrent processing in deep convolutional neural networks, but the computations underlying recurrent processing remain unclear. In this article, we tested a form of recurrence in deep residual networks (ResNets) to capture recurrent processing signals in the human brain. Although ResNets are feedforward networks, they approximate an excitatory additive form of recurrence. Essentially, this form of recurrence consists of repeating excitatory activations in response to a static stimulus. Here, we used ResNets of varying depths (reflecting varying levels of recurrent processing) to explain EEG activity within a visual masking paradigm. Sixty-two humans and 50 artificial agents (10 ResNet models of depths −4, 6, 10, 18, and 34) completed an object categorization task. We show that deeper networks explained more variance in brain activity compared with shallower networks. Furthermore, all ResNets captured differences in brain activity between unmasked and masked trials, with differences starting at ∼98 msec (from stimulus onset). These early differences indicated that EEG activity reflected “pure” feedforward signals only briefly (up to ∼98 msec). After ∼98 msec, deeper networks showed a significant increase in explained variance, which peaks at ∼200 msec, but only within unmasked trials, not masked trials. In summary, we provided clear evidence that excitatory additive recurrent processing in ResNets captures some of the recurrent processing in humans.

Published
2022
Full Text
View/download PDF

44. On the Necessity of Recurrent Processing during Object Recognition: It Depends on the Need for Scene Segmentation

Author

Seijdel, Noor, Loke, Jessica, van de Klundert, Ron, Meer, Matthew van der, Quispel, Eva, Gaal, Simon van, Haan, Edward H.F. de, Scholte, H. Steven, Seijdel, Noor, Loke, Jessica, van de Klundert, Ron, Meer, Matthew van der, Quispel, Eva, Gaal, Simon van, Haan, Edward H.F. de, and Scholte, H. Steven

Abstract

Although feedforward activity may suffice for recognizing objects in isolation, additional visual operations that aid object recognition might be needed for real-world scenes. One such additional operation is figure-ground segmentation, extracting the relevant features and locations of the target object while ignoring irrelevant features. In this study of 60 human participants (female and male), we show objects on backgrounds of increasing complexity to investigate whether recurrent computations are increasingly important for segmenting objects from more complex backgrounds. Three lines of evidence show that recurrent processing is critical for recognition of objects embedded in complex scenes. First, behavioral results indicated a greater reduction in performance after masking objects presented on more complex backgrounds, with the degree of impairment increasing with increasing background complexity. Second, electroencephalography (EEG) measurements showed clear differences in the evoked response potentials between conditions around time points beyond feedforward activity, and exploratory object decoding analyses based on the EEG signal indicated later decoding onsets for objects embedded in more complex backgrounds. Third, deep convolutional neural network performance confirmed this interpretation. Feedforward and less deep networks showed a higher degree of impairment in recognition for objects in complex backgrounds compared with recurrent and deeper networks. Together, these results support the notion that recurrent computations drive figure-ground segmentation of objects in complex scenes.

Published
2021
Full Text
View/download PDF

48. Feedforward and recurrent processing in scene segmentation: electroencephalography and functional magnetic resonance imaging

Author

Scholte, H. Steven, Jolij, Jacob, Fahrenfort, Johannes J., and Lamme, Victor A. F.

Subjects
Magnetic resonance imaging -- Methods, Electroencephalography -- Methods, Macaques -- Physiological aspects, Macaques -- Psychological aspects, Cognition in animals -- Research, Visual perception -- Physiological aspects, Health, Psychology and mental health
Published
2008

Catalog

Books, media, physical & digital resources
See catalog results