Theoretical Neuroscience | Laurent Perrinet

Artificial neural networks and machine learning applied to the understanding of biological vision

Thu, 05 Mar 2026 00:00:00 +0000

Practical work: https://github.com/laurentperrinet/2026-03_UE-neurosciences-computationnelles/

📖 See the full publication:

Laurent U Perrinet (2026). Working Memory with Polychronous Chains. arXiv preprint arXiv:2604.14096.

Foveated Retinotopy Improves Classification and Localization in CNNs

Mon, 23 Feb 2026 00:00:00 +0000

Foveated Retinotopy in CNNs. We represent Left an input image and how it is transformed by foveated retinotopy. We show below a representative reconstruction showing that it also acts as a cortical zoom on the image around the point of fixation. The transformed image is then fed to the ResNet deep learning architecture.

From falcons spotting prey to humans recognizing faces, the ability to rapidly process visual information depends on a foveated retinal organization that provides high-acuity central vision while preserving low-resolution peripheral vision. This organization is conserved along early visual pathways, yet remains under-explored in machine learning. Here, we examine the impact of embedding a foveated retinotopic transformation as a preprocessing layer on convolutional neural networks (CNNs) for image classification. By applying a log-polar mapping to off-the-shelf models and retraining them, we achieve comparable accuracy while improving robustness to scale and rotation. We demonstrate that this architecture is highly sensitive to shifts in the fixation point and that this sensitivity provides an effective proxy for defining saliency maps that facilitate object localization. Our results demonstrate that foveated retinotopy encodes prior geometric knowledge, providing a solution for visual searches and a meaningful classification robustness and localization trade-off. These findings provides a proof of concept in order to connect principles of biological vision with artificial networks, suggesting new, robust and efficient approaches for computer vision systems.

Foveated Retinotopy simulated by a log-polar map. We represent Left an input image with some geometrical objects and how it is transformed by the log-polar representation that implements foveated retinotopy. This shows that a rotation amounts to a translation on the polar axis (abscissa) and a zoom to a translation on the ordinates. We show right a representative reconstructionshowing that it also acts as a cortical zoom on the image around the point of fixation.

Foveal Retinotopy and Dual Pathways: A Computational Model for Active Visual Search

Fri, 10 Oct 2025 00:00:00 +0000

Abstract

This thesis investigates visual search through the lens of the dual visual pathways found in biological systems : the ventral (“what”) pathway, involved in object recognition, and the dorsal (“where”) pathway, responsible for spatial localisation and saccadic planning. Drawing from both neuroscience and computer vision, we propose a computational framework that integrates deep convolutional neural networks (DCNNs) within a biologically inspired architecture grounded in foveal retinotopy. As a proof of concept, prior work has demonstrated that incorporating saccadic planning improves digit categorisation performance in a controlled environment. Building upon this foundation, the primary objective of this thesis is to extend the computational framework to natural images in more ecologically valid settings. Our contributions are as follows : (1) We introduce a novel framework for training and evaluating DCNNs using semantically grounded, task-specific labels ; (2) We bridge the gap between artificial models and biological substrates by emphasizing the role of foveal retinotopy in robust object categorisation and precise localisation ; (3) We disentangle the interplay between categorisation and localisation by proposing a novel “localisation-frame” dataset, aimed at guiding the design of a biologically plausible dorsal stream model ; and (4) We present an initial model of the dorsal pathway, leveraging the new dataset to develop interpretable and efficient active vision systems—where interpretability is achieved through modular and spatially structured representations, and efficiency is reflected in reduced computational cost during inference with saccade planning. Overall, this thesis extends the dual-stream computational paradigm for visual search, contributes tools for explainable active vision, and offers a platform to explore hypotheses about functional specialisation in the human visual cortex.

Keywords

Visual search, Dual visual pathways, Deep Convolutional Neuronal, Network, Foveal retinotopy, Active vision

Résumé

Cette thèse étudie la recherche visuelle à travers le prisme des deux voies visuelles identifiées dans les systèmes biologiques : la voie ventrale, impliquée dans la reconnaissance des objets, et la voie dorsale, responsable de la localisation spatiale et de la planification des saccades. S’inspirant à la fois des neurosciences et de la vision artificielle, nous proposons un cadre computationnel intégrant des réseaux neuronal convolutifs profonds (DCNN) dans une architecture biologiquement plausible, fondée sur la rétinotopie fovéale. Des travaux antérieurs ont démontré que l’intégration de la planification des saccades améliorait les performances de catégorisation de chiffres dans un environnement contrôlé. S’appuyant sur cette base, l’objectif principal de cette thèse est d’étendre ce cadre théorique à des images naturelles dans des contextes plus écologiquement valides. Nos contributions sont les suivantes : (1) Nous proposons un nouveau cadre de travail pour l’entraînement et l’évaluation des DCNN, basé sur la sémantique sous-jacente aux labels initialement définis dans la communauté de la recherche computationnelle, ce qui permet de définir des tâches écologiques spécifiques ; (2) nous rapprochons les modèles artificiels des substrats biologiques en soulignant le rôle crucial de la retinotopie fovéales pour une catégorisation robuste et une localisation précise. (3) Nous approfondissons la connaissance de l’interaction entre la catégorisation et la localisation en proposant un ensemble de résultats structuré autour de cette relation, afin de guider la conception d’un modèle plausible de la voie dorsale ; (4) Enfin, en nous appuyant sur ces résultats, nous proposons une première modélisation de la voie dorsale visant à développer des systèmes de vision active à la fois interprétables, grâce à des représentations modulables et spatialement structurées, et efficaces, grâce à la planification de saccades permettant de réduire les coûts de calcul liés à l’inférence. Dans l’ensemble, cette thèse apporte plusieurs éléments : elle enrichit le modèle de vision artificielle des deux voies majeures impliquées dans la recherche visuelle, elle permet de développer des outils de vision active interprétables et elle fournit un cadre pour étudier les hypothèses biologiques relatives à la spécialisation fonctionnelle des aires cérébrales dédiées à la vision chez l’être humain.

Mots-clés

Recherche visuelle, Voie visuel ventrale, Voie visuel dorsale, Réseau neuronal convolutifs profonds, Rétinotopie fovéale, Vision active

Jean-Nicolas Jérémie (2025). Soutenance de Jean-Nicolas Jérémie "Foveal Retinotopy and Dual Pathways: A Computational Model for Active Visual Search".

DynTex: A Real-Time Generative Model of Dynamic Naturalistic Luminance Textures

Mon, 01 Sep 2025 00:00:00 +0000

🚀 Excited to share our new paper:

“DynTex: A real-time generative model of dynamic naturalistic luminance textures”

…now published in Journal of Vision!

🔹 Why it matters: Dynamic textures (e.g., fire, water, foliage) are everywhere, but modeling them in real-time has been a challenge. DynTex bridges this gap with a biologically inspired, efficient approach.

🔹 Key innovation: A generative model that captures the spatiotemporal statistics of natural scenes while running in real-time.

🔹 Applications: Computer vision, neuroscience, VR/AR, and more.📖

Read it here: https://doi.org/10.1167/jov.25.11.2

More on: https://laurentperrinet.github.io/publication/meso-25/

#DynamicTextures #ComputationalNeuroscience #ComputerVision #GenerativeModels #OpenScience

linkedin, mastodon, bluesky
The Motion Clouds stimuli were originally presented in the following paper (page links to other sresources)
Paula Sanz Leon, Ivo Vanzetta, Guillaume S Masson, Laurent U Perrinet (2012). Motion Clouds: Model-based stimulus synthesis of natural-like random textures for the study of motion perception. Journal of Neurophysiology.
PDF Cite DOI HAL URL arXiv Supp
examples of use: https://laurentperrinet.github.io/sciblog/categories/motionclouds.html

Artificial neural networks and machine learning applied to the understanding of biological vision

Mon, 26 May 2025 14:00:00 +0000

Master M4NC de l’institut NeuroMod, cours Prospective Innovation and Research.

Classification of Mental Workload Spatial Effects using Riemannian Manifold

Wed, 01 Jan 2025 00:00:00 +0000

This year at #CCN2025 we will be showcasing our research on the classification of Mental Workload 🥵 Spatial Effects using Riemannian Manifold.

📅 When: Wednesday, August 13, 1:00 – 4:00 pm 📍 Where: CCN 2025 Conference Venue, de Brug & E-Hall 📋 What: Poster B152

It leverages advanced mathematical techniques to better understand and classify mental workloads, offering new insights into cognitive processes and potential applications in various fields such as neuroscience, psychology, and human-computer interaction.
By utilizing Riemannian geometry, this research provides a robust framework for analyzing spatial effects in mental workload, paving the way for more accurate and efficient classification methods. This contribution not only advances our theoretical understanding but also has practical implications for improving mental workload assessment and management.

See you there! 🚀

https://laurentperrinet.github.io/publication/choplin-25-ccn/

👏 CNRS @cnrs@social.numerique.gouv.fr - Aix-Marseille University - ONERA, The French Aerospace Lab CNRS

#CCN2025 #Mental #Workload #MentalWorkload #Riemannian #Manifold

Links:

Kernel Heterogeneity Improves Sparseness of Natural Images Representations

Tue, 20 Aug 2024 00:00:00 +0000

5 minutes summary: https://hugoladret.github.io/publications/ladret_et_al_sparsecoding/
In a nutshell: We found that sparse coding of images (here extended in a convolutional framework) is improved when using kernels with heterogeneous precision in how they encode orientation information. This was confirmed by learning, but also by comparison with what is observed in the statistics of natural images and in our recordings from neurons in primary visual cortex.

Epistemic uncertainty in a CSC dictionary improves both sparseness and reconstruction performance. (a) Elements from dictionaries with fixed epistemic uncertainty before (green) and after dictionary learning (orange). (b) Elements from a dictionary with heterogeneous epistemic uncertainty before (blue) and after dictionary learning (purple). (c) Elements from a dictionary learned from scratch. (d) Distribution of the sparseness (top) and Peak Signal-to-Noise Ratio (PSNR, right) of the five dictionaries, shown as a scatter plot for each of the 600 images of the dataset (center). Median values are shown as dashed line on the histograms.
open access: https://iopscience.iop.org/article/10.1088/2634-4386/ad5d0f
This work is a followup of
Hugo Ladret, Laurent U Perrinet (2023). Convolutional Sparse Coding is improved by heterogeneous uncertainty modeling. ICLR 2023 SNN Workshop.
PDF Cite URL
This theoretical work accompanies a similar study in neurophysiology:
Hugo Ladret, Nelson Cortes, Lamyae Ikan, Frédéric Chavane, Christian Casanova, Laurent U Perrinet (2023). Cortical recurrence supports resilience to sensory variance in the primary visual cortex. Nature Communications Biology.
PDF Cite DOI Dataset Code URL bioRxiv HAL

Artificial neural networks and machine learning applied to the understanding of biological vision

Mon, 13 May 2024 14:00:00 +0000

Convolutional Sparse Coding is improved by heterogeneous uncertainty modeling

Fri, 05 May 2023 00:00:00 +0000

Accepted paper (poster) at the ICLR 2023 Workshop on Sparsity in Neural Networks:
the focus of the WS is on “On practical limitations and tradeoffs between sustainability and efficiency” in Kigali, Rwanda / May 5th 2023
reviews will be made public on https://openreview.net/forum?id=tgr8FEcl28M
In a nutshell: We found that sparse coding of images (here extended in a convolutional framework) is improved when using kernels with heterogeneous precision in how they encode orientation information. This was confirmed by learning, but also by comparison with what is observed in the statistics of natural images and in our recordings from neurons in primary visual cortex.

Epistemic uncertainty in a CSC dictionary improves both sparseness and reconstruction performance. (a) Elements from dictionaries with fixed epistemic uncertainty before (green) and after dictionary learning (orange). (b) Elements from a dictionary with heterogeneous epistemic uncertainty before (blue) and after dictionary learning (purple). (c) Elements from a dictionary learned from scratch. (d) Distribution of the sparseness (top) and Peak Signal-to-Noise Ratio (PSNR, right) of the five dictionaries, shown as a scatter plot for each of the 600 images of the dataset (center). Median values are shown as dashed line on the histograms.
This theoretical work accompanies a similar study in neurophysiology:
Hugo Ladret, Nelson Cortes, Lamyae Ikan, Frédéric Chavane, Christian Casanova, Laurent U Perrinet (2023). Cortical recurrence supports resilience to sensory variance in the primary visual cortex. Nature Communications Biology.
PDF Cite DOI Dataset Code URL bioRxiv HAL
This work was extended in
Hugo Ladret, Christian Casanova, Laurent U Perrinet (2024). Kernel Heterogeneity Improves Sparseness of Natural Images Representations. Neuromorphic Computing and Engineering.
Cite Dataset DOI URL HAL

Artificial neural networks and machine learning applied to the understanding of biological vision

Wed, 05 Apr 2023 09:00:00 +0000

Artificial neural networks and machine learning applied to the understanding of biological vision

Mon, 03 Apr 2023 14:00:00 +0000

Ultra-Fast Image Categorization in biology and in neural models

Tue, 21 Mar 2023 00:00:00 +0000

read the paper online or in PDF
full code with extensive Supplementary Material
Video Abstract and code for Video Abstract
join the Zotero group to add and discuss more items
this is a follow-up of:
Jean-Nicolas Jérémie, Laurent U Perrinet (2021). Ultra-fast categorization of images containing animals in vivo and in computo. Champalimaud Research Symposium (CRS21).
PDF Cite Venue
see an extension perspective to visual search in:
Jean-Nicolas Jérémie, Emmanuel Daucé, Laurent U Perrinet (2022). Ultra-rapid visual search in natural images using active deep learning. Proceedings of AREADNE.
PDF Cite Venue

Réseaux de neurones artificiels et apprentissage machine appliqués à la compréhension de la vision

Wed, 23 Mar 2022 09:00:00 +0000

Où: Salle PHY51 - Marseille (France)
Quoi: Master 1 Neurosciences et Sciences Cognitives

Réseaux neuronaux artificiels pour la vision

Mercredi 23/03/2022 de 9h-12h
Introduction aux Neurosciences de la Vision
Réseaux de neurones artificiels et apprentissage machine
slides

Neurones impulsionnels et modèles des fonctions visuelles

Mercredi 23/03/2022 de 13h30-16h30
TP via notebook
https://github.com/laurentperrinet/2022_UE-neurosciences-computationnelles/

Statistics of the sparse representations of natural images

Tue, 22 Mar 2022 00:00:00 +0000

see previous work: https://laurentperrinet.github.io/sciblog/posts/2018-11-05-statistics-of-the-natural-input-to-a-ring-model.html

Mini-Symposium “Learning from vision: Efficient representation, sparse coding, and modelling”

Although recent years have seen a striking improvement in imaging techniques, there are many tasks for which human interaction is still essential, as color gamut correction in the cinema industry. This suggests that a better understanding of the mechanisms underlying the visual system is instrumental to advances in imaging techniques. Along these lines, various ideas from computational neurosciences have found application in imaging, from pattern recognition to image inpainting. A promising line of investigation is built on methods based on models of the primary visual cortex and on neural coding, in particular via the efficient representation principle. These methods have recently allowed to define new artificial neural networks paradigms and to reproduce complex visual illusions. In this mini-symposium we aim to gather together experts working in the field of mathematical neuroscience and imaging, with a focus on these methods. In particular, the speakers will present recent results based on sparse coding and models of the visual system.

Organizer: Dario Prandi

12:40-1:05 The intrinsically nonlinear nature of receptive fields in vision: implications for imaging, vision science and artificial neural networks Marcelo Bertalmío, Spanish National Research Council, Spain
1:10-1:35 ChebLieNet: Invariant Spectral Graph Nns Turned Equivariant by Sub-Riemannian Geometry on Lie Groups Erik Bekkers, University of Amsterdam, Netherlands
1:40-2:05 Deep Predictive Coding for More Robust and Human-Like Vision Rufin VanRullen, Centre de Recherche Cerveau et Cognition (CerCo), France
2:10-2:35 Statistics of the Sparse Representations of Natural Images Hugo Ladret and Laurent U. Perrinet, CNRS & Aix-Marseille Université, Marseille, France More on https://meetings.siam.org/sess/dsp_programsess.cfm?sessioncode=73028

Dynamical processing of orientation precision in the primary visual cortex

Fri, 27 Aug 2021 00:00:00 +0000

This is 40th edition of Dynamicsdays
Nice, 23-27 August 2021 - https://dynamicsdays2021.univ-cotedazur.fr
check out the book of abstracts
In this talk, we will present the following paper :
Preliminary Program:
- Bruno Cessac, The Retina as a Dynamical System
- Hugo Ladret & Laurent Perrinet, Dynamics of the processing of orientation precision in the primary visual cortex
- Gianluigi Mongillo, Glassy phase in dynamically balanced networks
- Romain Veltz, Spatial and color hallucinations in a mathematical model of primary visual cortex

Pooling in a predictive model of V1 explains functional and structural diversity across species

Tue, 15 Jun 2021 11:15:00 +0000

In this talk, I will present the following paper :
Pooling in a predictive model of V1 explains functional and structural diversity across species
Victor Boutin, Angelo Franciosini, Frédéric Chavane, Laurent U Perrinet

Cite DOI URL bioRxiv
see a follow-up in:
Sparse Deep Predictive Coding captures contour integration capabilities of the early visual system
Victor Boutin, Angelo Franciosini, Frédéric Chavane, Franck Ruffier, Laurent U Perrinet

PDF Cite DOI Code URL arXiv
more about the role of top-down connections:
Effect of top-down connections in Hierarchical Sparse Coding
Victor Boutin, Angelo Franciosini, Franck Ruffier, Laurent U Perrinet

PDF Cite DOI URL arXiv

Understanding natural vision using deep predictive coding

Fri, 25 Sep 2020 15:45:00 +0000

What:: talk @ Séminaire à l’Institut de Recherche sur les Phénomènes Hors Équilibre (IRPHÉ)
Who:: Perrinet, Laurent U
Where: Marseille (France), see https://laurentperrinet.github.io/talk/2020-09-25-irphe
When: 25/09/2020, time: 15:45:00-16:30:00
What:
- Slides @ https://laurentperrinet.github.io/2020-09-25_IRPHE
- Code for slides @ https://github.com/laurentperrinet/2020-09-25_IRPHE/
- Abstract: Building models which efficiently process images is a great source of inspiration to better understand the processes which underly our visual perception. I will present some classical models stemming from the Machine Learning community and propose some extensions inspired by Nature. For instance, Sparse Coding (SC) is one of the most successful frameworks to model neural computations at the local scale in the visual cortex. It directly derives from the efficient coding hypothesis and could be thought of as a competitive mechanism that describes visual stimulus using the activity of a small fraction of neurons. At the structural scale of the ventral visual pathways, feedforward models of vision (CNNs in the terminology of deep learning) take into account neurophysiological observations and provide as of today the most successful framework for object recognition tasks. Nevertheless, these models do not leverage the high density of feedback and lateral interactions observed in the visual cortex. In particular, these connections are known to integrate contextual and attentional modulations to feedforward signals. The Predictive Coding (PC) theory has been proposed to model top-down and bottom-up interaction between cortical regions. We will here introduce a model combining Sparse Coding and Predictive Coding in a hierarchical and convolutional architecture. Our model, called Sparse Deep Predictive Coding (SDPC), was trained on several different databases including faces and natural images. We analyze the SPDC from a computational and a biological perspective and we combine neuroscientific evidence with machine learning methods to analyze the impact of recurrent processing at both the neural organization and representational levels. These results from the SDPC model additionally demonstrate that neuro-inspiration might be the right methodology to design more powerful and more robust computer vision algorithms.

Visual search as active inference

Mon, 14 Sep 2020 18:00:00 +0000

see proceedings paper:
Visual search as active inference
Emmanuel Daucé, Laurent U Perrinet

PDF Cite DOI Slides Pdf Code URL
What:: talk @ 1st International Workshop on Active Inference (IWAI 2020)
Who:: Emmanuel Daucé and Laurent Perrinet
Where: Ghent (Belgium), gone virtual, see https://laurentperrinet.github.io/talk/2020-09-14-iwai
When: 14/09/2020, time: 12:20:00-12:40:00
What:
- Slides @ https://laurentperrinet.github.io/2020-09-14_IWAI
- Code for slides @ https://github.com/laurentperrinet/2020-09-14_IWAI/
- Abstract: Visual search is an essential cognitive ability, offering a prototypical control problem to be addressed with Active Inference. Under a Naive Bayes assumption, the maximisation of the information gain objective is consistent with the separation of the visual sensory flow in two independent pathways, namely the “What” and the “Where” pathways. On the “What” side, the processing of the central part of the visual field (the fovea) provides the current interpretation of the scene, here the category of the target. On the “Where” side, the processing of the full visual field (at lower resolution) is expected to provide hints about future central foveal processing given the potential realisation of saccadic movements. A map of the classification accuracies, as obtained by such counterfactual saccades, defines a utility function on the motor space, whose maximal argument prescribes the next saccade. The comparison of the foveal and the peripheral predictions finally forms an estimate of the future information gain, providing a simple and resource-efficient way to implement information gain seeking policies in active vision. This dual-pathway information processing framework is found efficient on a synthetic visual search task and we show here quantitatively the role of the precision encoded within the accuracy map. More importantly, it is expected to draw connections toward a more general actor-critic principle in action selection, with the accuracy of the central processing taking the role of a value (or intrinsic reward) of the previous saccade.

From the retina to action: Understanding visual processing

Fri, 03 Apr 2020 16:00:00 +0000

2020-04_UE-neurosciences-computationnelles, matériel pour le cours de modélisation

Où: Marseille (France)
Quoi: Master Neurosciences et Sciences Cognitives
But de ce travail: lire un article scientifique, pouvoir le reproduire avec des simulations d’un neurone et afin d’améliorer sa compréhension.
Modalités: les étudiants s’organisent seuls, en binome ou en trinome pour fournir un mémoire sous forme de notebook complété à partir du modèle qui est fourni. Suivez les balises TODO dans le notebook pour vous guider dans cette rédaction. Les commentaires doivent être fait en français (ou en anglais si nécessaire) dans le notebook (n’oubliez-pas de sauver vos changements) et envoyé par e-mail à mailto:laurent.perrinet@univ-amu.fr une fois votre travail fini (de préférence avant le 31 avri).
Outils nécessaires: Jupyter, avec numpy et matplotlib. Ce sont des outils standard et qui sont facilement installables sur toute plateforme. Si vous avez des problèmes, me joindre par e-mail 👇

Humans adapt their anticipatory eye movements to the volatility of visual motion properties

Sun, 26 Jan 2020 00:00:00 +0000

“Humans adapt their anticipatory eye movements to the volatility of visual motion properties”

At what point should we become alarmed? When faced with changes in the environment, the sensory system provides an effective response.

The current health situation has shown us how abruptly our environment can change from one state to another, tragically illustrating the volatility we can face. To understand this notion of volatility, let’s take the case of a doctor who, among the patients he receives, usually diagnoses one out of ten cases of flu. Suddenly, he gets 5 out of 10 patients who test positive. Is this an unfortunate coincidence or are we now sure that there is a switch to a flu episode? Recent events have shown us how difficult it is to make a rational decision in times of uncertainty, and in particular to decide when to act. However, mathematical solutions exist that adapt our behavior by optimally combining the information explored recently with that exploited in the past. In an article published in PLoS Computational Biology, Pasturel, Montagnini and Perrinet show that our brain responds to changes in the sensory environment in the same way as this mathematical model.

By manipulating the probability bias of the presentation of a visual target on a screen, this experiment manipulates the volatility of the environment in a controlled way by introducing switches in the probability bias. These switches randomly change the bias among different degrees of probability (both left and right). At each trial, the bias then generates a realization, either left (L) or right (R). The target moves in blocks of 50 trials (1 to 50) and these realizations are the only ones to be observed, the evolution of the bias and its shifts remaining hidden from the observer. Compared to the floating average that is conventionally used, a mathematical model can be deduced as a predictive average that allows to better follow the dynamics of the probability bias. Thanks to psychophysical experiments, we have shown that observers preferentially follow the predictive mean, rather than the floating mean, both in explicit judgements (predictive betting) and, more surprisingly, in the anticipatory movements of the eyes that are carried out without the observers being aware of them.

These theoretical and experimental results show that in this realistic situation in which the context changes at random moments throughout the experiment, our sensory system adapts to volatility in an adaptive manner over the course of the trials. In particular, the experiments show in two behavioural experiments that humans adapt to volatility at the early sensorimotor level, through their anticipatory eye movements, but also at a higher cognitive level, through explicit evaluations. These results thus suggest that humans (and future artificial systems) can use much richer adaptation strategies than previously assumed. They provide a better understanding of how humans adapt to changing environments in order to make judgements or plan responses based on information that varies over time.

read the preprint (the official online publication or in PDF is wrongly typeset: the editors inverted the images of figures 2 & 3, while keeping the captions. Unfortunately, the policy of the journal is to issue a correction, but not to correct it. There is therefore no official correct version on the PLoS* website.)
get a )
Abstract
HTML
PDF
supplementary info : https://github.com/laurentperrinet/PasturelMontagniniPerrinet2020/blob/master/Pasturel_etal2020_PLoS-CB_SI.pdf
Communiqué de presse INSB-CNRS (en français)
code for paper: https://github.com/laurentperrinet/PasturelMontagniniPerrinet2020
code for framework: https://github.com/chloepasturel/AnticipatorySPEM
code for the Bayesian model: https://github.com/laurentperrinet/bayesianchangepoint
code for figures Figure 1, Figure 2, Figure 3, Figure 4, Figure 5
video abstract (and the code for generating the video abstract)
Notre papier avec Chloe Pasturel et @MontagniniAnna figure dans les faits marquants 2020 de la Société des Neurosciences! Voir aussi https://lejournal.cnrs.fr/nos-blogs/aux-frontieres-du-cerveau/les-faits-marquants-2020-de-la-societe-de-neurosciences :

From the retina to action: Dynamics of predictive processing in the visual system

Wed, 01 Jan 2020 00:00:00 +0000

Find the text at https://laurentperrinet.github.io/Perrinet20PredictiveProcessing/
The source code of the text is available at https://github.com/laurentperrinet/Perrinet20PredictiveProcessing This chapter is available as part of the book “The Philosophy and Science of Predictive Processing” : List of Contributors :
Preface: The Brain as a Prediction Machine, Anil Seth
Introduction, Dina Mendonça, Manuel Curado & Steven S. Gouveia
Part I: Predictive Processing: Philosophical Approaches

Predictive Processing and Representation: How Less Can Be More, Erik Myin and Thomas van Es
A Humean Challenge to Predictive Coding, Colin Klein
Are Markov Blankets Real and Does it Matter?, Richard Menary and Alexander J. Gillett
Predictive Processing and Metaphysical Views of the Self, Robert Clowes and Klaus Gärtner

Part II: Predictive Processing: Cognitive Science and Neuroscientific Approaches

From the Retina to Action: Dynamics of Predictive Processing in the Visual System, Laurent Perrinet
Predictive Processing and Consciousness: Prediction Fallacy and its Spatiotemporal Resolution, Steven S. Gouveia
The Many Faces of Attention: Why Precision Optimization is not Attention, Sina Fazelpour and Madeleine Ransom
Predictive Processing: Does it Compute?, Chris Thornton

Part III: Predictive Processing: Mental Health

The Predictive Brain, Conscious Experience and Brain-related Conditions, Lisa Feldman Barrett and Lorena Chanes
Disconnection and Diaschisis: Active Inference in Neuropsychology, Thomas Parr and Karl Friston
The Phenomenology and Predictive Processing of Time in Depression, Zachariah Neemeh and Shaun Gallagher
Why Use Predictive Processing to Explain Psychopathology? The Case of Anorexia Nervosa, Jakob Hohwy and Stephen Gadsby

Afterword, Manuel Curado

Learning where to look: a foveated visuomotor control model

Mon, 15 Jul 2019 12:20:00 +0000

download a preliminary PDF

Problem setting: In generic, ecological settings, the visual system faces a tricky problem when searching for one target (from a class of targets) in a cluttered environment. A) It is synthesized in the following experiment: After a fixation period of 200 ms, an observer is presented with a luminous display showing a single target from a known class (here digits) and at a random position. The display is presented for a short period of 500 ms (light shaded area in B), that is enough to perform at most one saccade (here, successful) on the potential target. Finally, the observer has to identify the digit by a keypress. B) Prototypical trace of a saccadic eye movement to the target position. In particular, we show the fixation window and the temporal window during which a saccade is possible (green shaded area). C) Simulated reconstruction of the visual information from the (interoceptive) retinotopic map at the onset of the display and after a saccade, the dashed red box indicating the visual area of the ``what’’ pathway. In contrast to an exteroceptive representation (see A), this demonstrates that the position of the target has to be inferred from a degraded (sampled) image. In particular, the configuration of the display is such that by adding clutter and reducing the size of the digit, it may become necessary to perform a saccade to be able to identify the digit. The computational pathway mediating the action has to infer the location of the target \emph{before seeing it}, that is, before being able to actually identify the target’s category from a central fixation.

Results: success

Results: failure to classify

Results: failure to locate

From the retina to action: Understanding visual processing

Wed, 03 Apr 2019 16:00:00 +0000

Cours de Licence Sciences & Humanité, 3/4/2019

From the retina to action: Predictive processing in the visual system

Mon, 25 Mar 2019 14:30:00 +0000

see a follow-up in:
Sparse Deep Predictive Coding captures contour integration capabilities of the early visual system
Victor Boutin, Angelo Franciosini, Frédéric Chavane, Franck Ruffier, Laurent U Perrinet

PDF Cite DOI Code URL arXiv
more about the role of top-down connections:
Effect of top-down connections in Hierarchical Sparse Coding
Victor Boutin, Angelo Franciosini, Franck Ruffier, Laurent U Perrinet

PDF Cite DOI URL arXiv

An adaptive homeostatic algorithm for the unsupervised learning of visual features

Tue, 01 Jan 2019 00:00:00 +0000

“An adaptive algorithm for unsupervised learning”

supplementary info : https://spikeai.github.io/HULK/
Abstract
HTML
PDF
code for paper: https://github.com/SpikeAI/HULK
code for framework: https://github.com/bicv/SparseHebbianLearning/
code for figures https://github.com/SpikeAI/HULK/blob/master/Annex.ipynb (which is rendered @ https://spikeai.github.io/HULK/ )
video abstract (and the code for generating it)
previous publication :
Laurent U Perrinet (2010). Role of homeostasis in learning sparse representations. Neural Computation.
PDF Cite DOI Hal Code URL arXiv

$$f(k;p_{0}^{*}) = \begin{cases}p_{0}^{*} & \text{if }k=1, \\ 1-p_{0}^{*} & \text{if }k=0.\end{cases}$$

Bayesian Modeling of Motion Perception using Dynamical Stochastic Textures

Wed, 21 Nov 2018 00:00:00 +0000

The flash-lag effect as a motion-based predictive shift

Thu, 26 Jan 2017 00:00:00 +0000

Press release

Visual illusions: their origin lies in prediction

Flash-Lag Effect. When a visual stimulus moves along a continuous trajectory, it may be seen ahead of its veridical position with respect to an unpredictable event such as a punctuate flash. This illusion tells us something important about the visual system: contrary to classical computers, neural activity travels at a relatively slow speed. It is largely accepted that the resulting delays cause this perceived spatial lag of the flash. Still, after several decades of debates, there is no consensus regarding the underlying mechanisms.

Researchers from the Timone Institute of Neurosciences bring a new theoretical hypothesis on a visual illusion discovered at the beginning of the 20th century. This illusion remained misunderstood while it poses fundamental questions about how our brains represent events in space and time. This study published on January 26, 2017 in the journal PLOS Computational Biology, shows that the solution lies in the predictive mechanisms intrinsic to the neural processing of information.

Visual illusions are still popular: in a quasi-magical way, they can make objects appear where they are not expected… They are also excellent opportunities to question the constraints of our perceptual system. Many illusions are based on motion, such as the flash-lag effect. Observe a luminous dot that moves along a rectilinear trajectory. If a second light dot is flashed very briefly just above the first, the moving point will always be perceived in front of the flash while they are vertically aligned.

Fig 2. Diagonal Markov chain. In the current study, the estimated state vector z = {x, y, u, v} is composed of the 2D position (x and y) and velocity (u and v) of a (moving) stimulus. (A) First, we extend a classical Markov chain using Nijhawan’s diagonal model in order to take into account the known neural delay τ: At time t, information is integrated until time t − τ, using a Markov chain and a model of state transitions p(zt|zt−δt) such that one can infer the state until the last accessible information p(zt−τ|I0:t−τ). This information can then be “pushed” forward in time by predicting its trajectory from t − τ to t. In particular p(zt|I0:t−τ) can be predicted by the same internal model by using the state transition at the time scale of the delay, that is, p(zt|zt−τ). This is virtually equivalent to a motion extrapolation model but without sensory measurements during the time window between t − τ and t. Note that both predictions in this model are based on the same model of state transitions. (B) One can write a second, equivalent “pull” mode for the diagonal model. Now, the current state is directly estimated based on a Markov chain on the sequence of delayed estimations. While being equivalent to the push-mode described above, such a direct computation allows to more easily combine information from areas with different delays. Such a model implements Nijhawan’s “diagonal model”, but now motion information is probabilistic and therefore, inferred motion may be modulated by the respective precisions of the sensory and internal representations. (C) Such a diagonal delay compensation can be demonstrated in a two-layered neural network including a source (input) and a target (predictive) layer [44]. The source layer receives the delayed sensory information and encodes both position and velocity topographically within the different retinotopic maps of each layer. For the sake of simplicity, we illustrate only one 2D map of the motions (x, v). The integration of coherent information can either be done in the source layer (push mode) or in the target layer (pull mode). Crucially, to implement a delay compensation in this motion-based prediction model, one may simply connect each source neuron to a predictive neuron corresponding to the corrected position of stimulus (x + v ⋅ τ, v) in the target layer. The precision of this anisotropic connectivity map can be tuned by the width of convergence from the source to the target populations. Using such a simple mapping, we have previously shown that the neuronal population activity can infer the current position along the trajectory despite the existence of neural delays.

Processing visual information takes time and even if these delays are remarkably short, they are not negligible and the nervous system must compensate them. For an object that moves predictably, the neural network can infer its most probable position taking into account this processing time. For the flash, however, this prediction can not be established because its appearance is unpredictable. Thus, while the two targets are aligned on the retina at the time of the flash, the position of the moving object is anticipated by the brain to compensate for the processing time: it is this differentiated treatment that causes the flash-lag effect. The researchers show that this hypothesis also makes it possible to explain the cases where this illusion does not work: for example if the flash appears at the end of the moving dot’s trajectory or if the target reverses its path in an unexpected way. In this work, the major innovation is to use the accuracy of information in the dynamics of the model. Thus, the corrected position of the moving target is calculated by combining the sensory flux with the internal representation of the trajectory, both of which exist in the form of probability distributions. To manipulate the trajectory is to change the precision and therefore the relative weight of these two information when they are optimally combined in order to know where an object is at the present time. The researchers propose to call parodiction (from the ancient Greek paron, the present) this new theory that joins Bayesian inference with taking into account neuronal delays.

Fig 5. Histogram of the estimated positions as a function of time for the dMBP model. Histograms of the inferred horizontal positions (blueish bottom panel) and horizontal velocity (reddish top panel), as a function of time frame, from the dMBP model. Darker levels correspond to higher probabilities, while a light color corresponds to an unlikely estimation. We highlight three successive epochs along the trajectory, corresponding to the flash initiated, standard (mid-point) and flash terminated cycles. The timing of the flashes are respectively indicated by the dashed vertical lines. In dark, the physical time and in green the delayed input knowing τ = 100 ms. Histograms are plotted at two different levels of our model in the push mode. The left-hand column illustrates the source layer that corresponds to the integration of delayed sensory information, including the prior on motion. The right-hand illustrates the target layer corresponding to the same information but after the occurrence of some motion extrapolation compensating for the known neural delay τ.

Despite the simplicity of this solution, parodiction has elements that may seem counter-intuitive. Indeed, in this model, the physical world is considered “hidden”, that is to say, it can only be guessed by our sensations and our experience. The role of visual perception is then to deliver to our central nervous system the most likely information despite the different sources of noise, ambiguity and time delays. According to the authors of this publication, the visual treatment would consist in a “simulation” of the visual world projected at the present time, even before the visual information can actually modulate, confirm or cancel this simulation. This hypothesis, which seems to belong to “science fiction”, is being tested with more detailed and biologically plausible hierarchical neural network models that should allow us to better understand the mysteries underlying our perception. Visual illusions have still the power to amaze us!

check_out further results on introducing anisotropies in the FLE

Modelling the dynamics of cognitive processes: from the Bayesian brain to particles

Thu, 07 Jul 2016 13:00:00 +0000

See a followup in Perrinet et al, 2012

Sparse Models for Computer Vision

Sun, 01 Nov 2015 00:00:00 +0000

Appeared in this book:
Gabriel Cristóbal, Laurent U Perrinet, Matthias S Keil (2015). Biologically Inspired Computer Vision.
Cite DOI Code URL

Anticipatory smooth eye movements and reinforcement

Tue, 01 Sep 2015 00:00:00 +0000

A Mathematical Account of Dynamic Texture Synthesis for Probing Visual Perception

Thu, 01 Jan 2015 00:00:00 +0000

See a followup in
Jonathan Vacher, Andrew Isaac Meso, Laurent U Perrinet, Gabriel Peyré (2018). Bayesian Modeling of Motion Perception using Dynamical Stochastic Textures. Neural Computation.
PDF Cite DOI URL arXiv

Motion-based prediction model for flash lag effect

Fri, 22 Aug 2014 00:00:00 +0000

Based on
Laurent U Perrinet, Guillaume S Masson (2012). Motion-based prediction is sufficient to solve the aperture problem. Neural Computation.
PDF Cite Pdf arXiv Doi
see follow-up on motion extrapolation:
Mina A Khoei, Guillaume S Masson, Laurent U Perrinet (2013). Motion-based prediction explains the role of tracking in motion extrapolation. Journal of Physiology-Paris.
PDF Cite DOI URL
see follow-up on the flash-lag effect:
Mina A Khoei, Guillaume S Masson, Laurent U Perrinet (2017). The flash-lag effect as a motion-based predictive shift. PLoS Computational Biology.
Cite Slides DOI Press Code URL HAL

Signature of an anticipatory response in area V1 as modeled by a probabilistic model and a spiking neural network

Sun, 06 Jul 2014 00:00:00 +0000

Based on
Laurent U Perrinet, Guillaume S Masson (2012). Motion-based prediction is sufficient to solve the aperture problem. Neural Computation.
PDF Cite Pdf arXiv Doi
see follow-up on motion extrapolation:
Mina A Khoei, Guillaume S Masson, Laurent U Perrinet (2013). Motion-based prediction explains the role of tracking in motion extrapolation. Journal of Physiology-Paris.
PDF Cite DOI URL
see follow-up on the flash-lag effect:
Mina A Khoei, Guillaume S Masson, Laurent U Perrinet (2017). The flash-lag effect as a motion-based predictive shift. PLoS Computational Biology.
Cite Slides DOI Press Code URL HAL

Figure 4: Rasterplot of input and output spikes. The raster plot from excitatory neurons is ordered according to their position. Each input spike is a blue dot and each output spike is a black dot. While input is scattered during blanking periods (Figure 1), the network output shows shows some tuned activity during the blank (compare with the activity before visual stimulation). To decode such patterns of activity we used a maximum-likelihood estimation technique based on the tuning curve of the neurons.

WP5 - Demo 1.3 : Spiking model of motion-based prediction

Thu, 20 Mar 2014 13:00:00 +0000

Advances in Texture Analysis for Emphysema Classification

Tue, 01 Jan 2013 00:00:00 +0000

relies on log-Gabor filters:
Sylvain Fischer, Filip Šroubek, Laurent U Perrinet, Rafael Redondo, Gabriel Cristóbal (2007). Self-Invertible 2D Log-Gabor Wavelets. International Journal of Computer Vision.
PDF Cite DOI Code URL

Motion Clouds: Model-based stimulus synthesis of natural-like random textures for the study of motion perception

Wed, 14 Mar 2012 00:00:00 +0000

MotionClouds are random dynamic stimuli optimized to study motion perception.

Web-site
Source code using Python.
37 citations on Google Scholar (last updated 22/10/2021)
Supplementary information
Follow-up paper
Jonathan Vacher, Andrew Isaac Meso, Laurent U Perrinet, Gabriel Peyré (2018). Bayesian Modeling of Motion Perception using Dynamical Stochastic Textures. Neural Computation.
PDF Cite DOI URL arXiv

Jonathan Vacher, Andrew Isaac Meso, Laurent U Perrinet, Gabriel Peyré (2015). Biologically Inspired Dynamic Textures for Probing Motion Perception. Advances in Neural Information Processing Systems.
Cite Pdf PDF arXiv
This library was notably used in the following paper
Claudio Simoncini, Laurent U Perrinet, Anna Montagnini, Pascal Mamassian, Guillaume S Masson (2012). More is not always better: dissociation between perception and action explained by adaptive gain control. Nature Neuroscience.
PDF Cite DOI URL

Figure 4 Broadband vs. narrowband stimuli. From A through B to C, the frequency bandwidth Bf increases, while all other parameters (such as f0) are kept constant. The MC with the broadest bandwidth is thought to best represent natural stimuli, since, as those, it contains many frequency components. A: Bf = 0:05 (Supplemental Movie S4). B: Bf = 0:15 (Supplemental Movie S5). C: Bf = 0:4 (Supplemental Movie S6).

Active inference, smooth pursuit and oculomotor delays

Sun, 01 Jan 2012 00:00:00 +0000

Pursuing motion illusions: a realistic oculomotor framework for Bayesian inference

Fri, 22 Apr 2011 00:00:00 +0000

See a followup in
Laurent U Perrinet, Guillaume S Masson (2012). Motion-based prediction is sufficient to solve the aperture problem. Neural Computation.
PDF Cite Pdf arXiv Doi

Saccadic foveation of a moving visual target in the rhesus monkey

Tue, 01 Feb 2011 00:00:00 +0000

Probabilistic models of the low-level visual system: the role of prediction in detecting motion

Fri, 17 Dec 2010 00:00:00 +0000

An event ranging “From Mathematical Image Analysis to Neurogeometry of the Brain” Ladislav Tauc & GDR MSPC neurosciences conference.

related publication from Mina Khoei @ TAUC 2012
see this more recent talk @ UCL, London

Probabilistic models of the low-level visual system: the role of prediction in detecting motion

Fri, 01 Jan 2010 00:00:00 +0000

See a followup in Perrinet et al, 2012

Reading out the dynamics of lateral interactions in the primary visual cortex from VSD data

Mon, 30 Nov 2009 00:00:00 +0000

see this more recent poster @ VSS

Decoding center-surround interactions in population of neurons for the ocular following response

Thu, 01 Jan 2009 00:00:00 +0000

Inferring monkey ocular following responses from V1 population dynamics using a probabilistic model of motion integration

Thu, 01 Jan 2009 00:00:00 +0000

Decoding the population dynamics underlying ocular following response using a probabilistic framework

Sun, 01 Jun 2008 00:00:00 +0000

related publications @ FENS 2006, @ NeuroComp 2008 and @ AREADNE 2008
see this more recent talk @ UCL, London

Adaptive Sparse Spike Coding : applications of Neuroscience to the compression of natural images

Tue, 01 Jan 2008 00:00:00 +0000

Decoding the population dynamics underlying ocular following response using a probabilistic framework

Tue, 01 Jan 2008 00:00:00 +0000

Modeling spatial integration in the ocular following response to center-surround stimulation using a probabilistic framework

Tue, 01 Jan 2008 00:00:00 +0000

Neural Codes for Adaptive Sparse Representations of Natural Images

Sat, 01 Sep 2007 00:00:00 +0000

Dynamical Neural Networks: modeling low-level vision at short latencies

Thu, 01 Mar 2007 00:00:00 +0000

Dynamical Neural Networks (DyNNs) are a class of models for networks of neurons where particular focus is put on the role of time in the emergence of functional computational properties. The definition and study of these models involves the cooperation of a large range of scientific fields from statistical physics, probabilistic modelling, neuroscience and psychology to control theory. It focuses on the mechanisms that may be relevant for studying cognition by hypothesizing that information is distributed in the activity of the neurons in the system and that the timing helps in maintaining this information to lastly form decisions or actions. The system responds at best to the constraints of the outside world and learning strategies tune this internal dynamics to achieve optimal performance. This chapter introduces the book. See also:

Bruno Cessac, Emmanuel Daucé, Laurent U Perrinet, Manuel Samuelides (2007). Topics in Dynamical Neural Networks: From Large Scale Neural Networks to Motor Control and Vision. Topics in Dynamical Neural Networks: From Large Scale Neural Networks to Motor Control and Vision.
Cite
Bruno Cessac, Emmanuel Daucé, Laurent U Perrinet, Manuel Samuelides (2007). Introduction to Topics in Dynamical Neural Networks: From Large Scale Neural Networks to Motor Control and Vision. Topics in Dynamical Neural Networks: From Large Scale Neural Networks to Motor Control and Vision.
Cite DOI URL

Self-Invertible 2D Log-Gabor Wavelets

Sat, 13 Jan 2007 00:00:00 +0000

This library defines the set of LogGabor kernels. These are generic edge-like filters at different scales, phases and orientations. The library develops a simple method to construct a simple multi-scale linear transform.

Web-site
Source code
logGabor filters are used in numerous computer vision applications and reaches 177 citations on Google Scholar (last updated 22/10/2021).

Figure 1 Multiresolution schemes. (a) Schematic contours of the log-Gabor filters in the Fourier domain with 5 scales and 8 orientations (only the contours at 78% of the filter maximum are drawn). (b) The real part of the corresponding filters is drawn in the spatial domain. The two first scales are drawn at the bottom magnified by a factor of 4 for a better visualization. The different scales are arranged in rows and the orientations in columns. The low-pass filter is drawn in the upper-left part. (c) The corresponding imaginary parts of the filters are shown in the same arrangement. Note that the low-pass filter does not have imaginary part. Insets (b) and (c) show the final filters built through all the processes described in Section 2. (d) In the proposed scheme the elongation of log-Gabor wavelets increases with the number of orientations nt . Here the real parts (left column) and imaginary parts (right column) are drawn for the 3, 4, 6, 8, 10, 12 and 16 orientation schemes. (e) As a comparison orthogonal wavelet filters ‘Db4’ are shown. Horizontal, vertical and diagonal wavelets are arranged on columns (low-pass on top). (f) As a second comparison, steerable pyramid filters (Portilla et al., 2003) are shown. The arrangement over scales and orientations is the same as for the log-Gabor scheme.

Bayesian modeling of dynamic motion integration

Mon, 01 Jan 2007 00:00:00 +0000

Modeling spatial integration in the ocular following response using a probabilistic framework

Mon, 01 Jan 2007 00:00:00 +0000

Neural Codes for Adaptive Sparse Representations of Natural Images

Mon, 01 Jan 2007 00:00:00 +0000

Bayesian modeling of dynamic motion integration

Sun, 01 Jan 2006 00:00:00 +0000

See a followup in Perrinet et al, 2012

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Sun, 01 Jan 2006 00:00:00 +0000

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Sun, 01 Jan 2006 00:00:00 +0000

See a followup in Perrinet et al, 2012

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Sun, 01 Jan 2006 00:00:00 +0000

related publication @ SPIE 2008
See a followup in Perrinet et al, 2012

Coding static natural images using spiking event times: do neurons cooperate?

Wed, 01 Sep 2004 00:00:00 +0000

Progressive reconstruction of a static image using spikes in a multi-scale oriented representation.

Finding Independent Components using spikes : a natural result of Hebbian learning in a sparse spike coding scheme

Thu, 01 Jan 2004 00:00:00 +0000

Network of integrate-and-fire neurons using Rank Order Coding B: spike timing dependant plasticity and emergence of orientation selectivity

Mon, 01 Jan 2001 00:00:00 +0000

Theoretical Neuroscience | Laurent Perrinet

Artificial neural networks and machine learning applied to the understanding of biological vision

Foveated Retinotopy Improves Classification and Localization in CNNs

links

Foveal Retinotopy and Dual Pathways: A Computational Model for Active Visual Search

Abstract

Keywords

Résumé

Mots-clés

DynTex: A Real-Time Generative Model of Dynamic Naturalistic Luminance Textures

Artificial neural networks and machine learning applied to the understanding of biological vision

Master M4NC de l’institut NeuroMod, cours Prospective Innovation and Research.

Classification of Mental Workload Spatial Effects using Riemannian Manifold

Kernel Heterogeneity Improves Sparseness of Natural Images Representations

Artificial neural networks and machine learning applied to the understanding of biological vision

Convolutional Sparse Coding is improved by heterogeneous uncertainty modeling

Artificial neural networks and machine learning applied to the understanding of biological vision

Artificial neural networks and machine learning applied to the understanding of biological vision

Ultra-Fast Image Categorization in biology and in neural models

Réseaux de neurones artificiels et apprentissage machine appliqués à la compréhension de la vision

Statistics of the sparse representations of natural images

Mini-Symposium “Learning from vision: Efficient representation, sparse coding, and modelling”

Organizer: Dario Prandi

Dynamical processing of orientation precision in the primary visual cortex

Pooling in a predictive model of V1 explains functional and structural diversity across species

Understanding natural vision using deep predictive coding

Visual search as active inference

From the retina to action: Understanding visual processing

2020-04_UE-neurosciences-computationnelles, matériel pour le cours de modélisation

Humans adapt their anticipatory eye movements to the volatility of visual motion properties

“Humans adapt their anticipatory eye movements to the volatility of visual motion properties”

At what point should we become alarmed? When faced with changes in the environment, the sensory system provides an effective response.

From the retina to action: Dynamics of predictive processing in the visual system

Learning where to look: a foveated visuomotor control model

From the retina to action: Understanding visual processing

From the retina to action: Predictive processing in the visual system

An adaptive homeostatic algorithm for the unsupervised learning of visual features

“An adaptive algorithm for unsupervised learning”

Bayesian Modeling of Motion Perception using Dynamical Stochastic Textures

The flash-lag effect as a motion-based predictive shift

Visual illusions: their origin lies in prediction

Modelling the dynamics of cognitive processes: from the Bayesian brain to particles

Sparse Models for Computer Vision

Anticipatory smooth eye movements and reinforcement

A Mathematical Account of Dynamic Texture Synthesis for Probing Visual Perception

Motion-based prediction model for flash lag effect

Signature of an anticipatory response in area V1 as modeled by a probabilistic model and a spiking neural network

WP5 - Demo 1.3 : Spiking model of motion-based prediction

Advances in Texture Analysis for Emphysema Classification

Motion Clouds: Model-based stimulus synthesis of natural-like random textures for the study of motion perception

Active inference, smooth pursuit and oculomotor delays

Pursuing motion illusions: a realistic oculomotor framework for Bayesian inference

Saccadic foveation of a moving visual target in the rhesus monkey

Probabilistic models of the low-level visual system: the role of prediction in detecting motion

Probabilistic models of the low-level visual system: the role of prediction in detecting motion

Reading out the dynamics of lateral interactions in the primary visual cortex from VSD data

Decoding center-surround interactions in population of neurons for the ocular following response

Inferring monkey ocular following responses from V1 population dynamics using a probabilistic model of motion integration

Decoding the population dynamics underlying ocular following response using a probabilistic framework

Adaptive Sparse Spike Coding : applications of Neuroscience to the compression of natural images

Decoding the population dynamics underlying ocular following response using a probabilistic framework

Modeling spatial integration in the ocular following response to center-surround stimulation using a probabilistic framework

Neural Codes for Adaptive Sparse Representations of Natural Images

Dynamical Neural Networks: modeling low-level vision at short latencies

Self-Invertible 2D Log-Gabor Wavelets

Bayesian modeling of dynamic motion integration

Modeling spatial integration in the ocular following response using a probabilistic framework

Neural Codes for Adaptive Sparse Representations of Natural Images

Bayesian modeling of dynamic motion integration

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Input-output transformation in the visuo-oculomotor loop: modeling the ocular following response to center-surround stimulation in a probabilistic framework

Coding static natural images using spiking event times: do neurons cooperate?

Finding Independent Components using spikes : a natural result of Hebbian learning in a sparse spike coding scheme

Network of integrate-and-fire neurons using Rank Order Coding B: spike timing dependant plasticity and emergence of orientation selectivity