scholarly journals Convis: A Toolbox To Fit and Simulate Filter-based Models of Early Visual Processing

2017 ◽  
Author(s):  
Jacob Huth ◽  
Timothée Masquelier ◽  
Angelo Arleo
Keyword(s):  
Visual Processing ◽  
Large Scale ◽  
Automatic Differentiation ◽  
Receptive Fields ◽  
Gain Control ◽  
Simulation Software ◽  
Linear Filters ◽  
Contrast Gain ◽  
Selective Filter ◽  
Link Type

1AbstractWe developed Convis, a Python simulation toolbox for large scale neural populations which offers arbitrary receptive fields by 3D convolutions executed on a graphics card. The resulting software proves to be flexible and easily extensible in Python, while building on the PyTorch library [32], which was previously used successfully in deep learning applications, for just-in-time optimization and compilation of the model onto CPU or GPU architectures. An alternative implementation based on Theano [33] is also available, although not fully supported.Through automatic differentiation, any parameter of a specified model can be optimized to approach a desired output which is a significant improvement over e.g. Monte Carlo or particle optimizations without gradients. We show that a number of models including even complex non-linearities such as contrast gain control and spiking mechanisms can be implemented easily.We show in this paper that we can in particular recreate the simulation results of a popular retina simulation software VirtualRetina [35], with the added benefit of providing (1) arbitrary linear filters instead of the product of Gaussian and exponential filters and (2) optimization routines utilizing the gradients of the model. We demonstrate the utility of 3d convolution filters with a simple direction selective filter. Also we show that it is possible to optimize the input for a certain goal, rather than the parameters, which can aid the design of experiments as well as closed-loop online stimulus generation. Yet, Convis is more than a retina simulator. For instance it can also predict the response of V1 orientation selective cells.Convis is open source under the GPL-3.0 license and available from https://github.com/jahuth/convis/ with documentation at https://jahuth.github.io/convis/.

scholarly journals A Differential Model of the Complex Cell

2011 ◽  
Vol 23 (9) ◽  
pp. 2324-2357 ◽  
Author(s):  
Miles Hansard ◽  
Radu Horaud
Keyword(s):  
Visual Processing ◽  
Receptive Fields ◽  
Scale Space ◽  
Two Dimensions ◽  
Complex Cell ◽  
Linear Filters ◽  
Differential Model ◽  
New Model ◽  
Statistical Measures ◽  
Gabor Functions

The receptive fields of simple cells in the visual cortex can be understood as linear filters. These filters can be modeled by Gabor functions or gaussian derivatives. Gabor functions can also be combined in an energy model of the complex cell response. This letter proposes an alternative model of the complex cell, based on gaussian derivatives. It is most important to account for the insensitivity of the complex response to small shifts of the image. The new model uses a linear combination of the first few derivative filters, at a single position, to approximate the first derivative filter, at a series of adjacent positions. The maximum response, over all positions, gives a signal that is insensitive to small shifts of the image. This model, unlike previous approaches, is based on the scale space theory of visual processing. In particular, the complex cell is built from filters that respond to the 2D differential structure of the image. The computational aspects of the new model are studied in one and two dimensions, using the steerability of the gaussian derivatives. The response of the model to basic images, such as edges and gratings, is derived formally. The response to natural images is also evaluated, using statistical measures of shift insensitivity. The neural implementation and predictions of the model are discussed.

scholarly journals Topographic organization of the “third tier” dorsomedial visual cortex in the macaque

2019 ◽  
Author(s):  
Kostas Hadjidimitrakis ◽  
Sophia Bakola ◽  
Tristan A. Chaplin ◽  
Hsin-Hao Yu ◽  
Omar Alanazi ◽  
...  
Keyword(s):  
Visual Cortex ◽  
Visual Field ◽  
Visual Processing ◽  
Large Scale ◽  
Receptive Fields ◽  
Topographic Map ◽  
Lower Quadrant ◽  
Functional Studies ◽  
The Third ◽  
Upper Visual Field

AbstractThe boundaries of the visual areas located anterior to V2 in the dorsomedial region of the macaque cortex remain contentious. This region is usually conceptualized as including two functional subdivisions: the dorsal component of area V3 (V3d), laterally, and another area, named the parietooccipital area (PO) or V6, medially. However, the nature of the putative border between V3d and PO/V6 has remained undefined. We recorded the receptive fields of multiunit clusters in adult male macaques, and reconstructed the locations of recording sites using histological sections and “unfolded” cortical maps. Immediately adjacent to dorsomedial V2 we observed a representation of the lower contralateral quadrant, which represented the vertical meridian at its rostral border. This region, corresponding to V3d of previous studies, formed a simple eccentricity gradient, from approximately <5° in the annectant gyrus, to >60° in the parietooccipital sulcus. However, there was no topographic reversal where one would expect to find the border between V3d and PO/V6. Rather, near the midline, this lower quadrant map continued directly into a representation of the peripheral upper visual field, without an intervening lower quadrant representation that could be unambiguously assigned to PO/V6. Thus, V3d and PO/V6 form a continuous topographic map, which includes parts of both quadrants. Together with previous observations that V3d and PO/V6 are both densely myelinated relative to adjacent cortex, and share similar input from V1, these results suggest that they are parts of a single area, which is distinct from the one forming the ventral component of the third tier complex.Significance statementThe primate visual cortex has a large number of areas. Knowing the extent of each visual area, and how they can be distinguished from each other, are essential for the interpretation of experiments aimed at understanding visual processing. Currently, there are conflicting models of the organization of the dorsomedial visual cortex rostral to area V2 (one of the earliest stages of cortical processing of vision). By conducting large-scale electrophysiological recordings, we found that what were originally thought to be distinct areas in this region (dorsal V3, and the parietooccipital area [PO/V6]), together form a single map the visual field. These results will help guide future functional studies, and the interpretation of the outcomes of lesions involving the dorsal visual cortex.

scholarly journals Proteomic profiling dataset of chemical perturbations in multiple biological backgrounds

2021 ◽  
Vol 8 (1) ◽  
Author(s):  
Deborah O. Dele-Oni ◽  
Karen E. Christianson ◽  
Shawn B. Egri ◽  
Alvaro Sebastian Vaca Jacome ◽  
Katherine C. DeRuff ◽  
...  
Keyword(s):  
Large Scale ◽  
Cell Model ◽  
Cellular Responses ◽  
Proteomic Profiling ◽  
Reduced Representation ◽  
Link Type ◽  
Original Dataset ◽  
Quality Control Metrics ◽  
Biological Insight ◽  
Chromatin Profiling

AbstractWhile gene expression profiling has traditionally been the method of choice for large-scale perturbational profiling studies, proteomics has emerged as an effective tool in this context for directly monitoring cellular responses to perturbations. We previously reported a pilot library containing 3400 profiles of multiple perturbations across diverse cellular backgrounds in the reduced-representation phosphoproteome (P100) and chromatin space (Global Chromatin Profiling, GCP). Here, we expand our original dataset to include profiles from a new set of cardiotoxic compounds and from astrocytes, an additional neural cell model, totaling 5300 proteomic signatures. We describe filtering criteria and quality control metrics used to assess and validate the technical quality and reproducibility of our data. To demonstrate the power of the library, we present two case studies where data is queried using the concept of “connectivity” to obtain biological insight. All data presented in this study have been deposited to the ProteomeXchange Consortium with identifiers PXD017458 (P100) and PXD017459 (GCP) and can be queried at https://clue.io/proteomics.

The Finite Element Analysis of the Large-Scale Converter Transformer Valve Side of the Electric Field

2013 ◽  
Vol 325-326 ◽  
pp. 476-479 ◽  
Author(s):  
Lin Suo Zeng ◽  
Zhe Wu
Keyword(s):  
Finite Element ◽  
Electric Field ◽  
Large Scale ◽  
Calculation Model ◽  
Simulation Software ◽  
Field Calculation ◽  
Element Analysis ◽  
Ansys Simulation ◽  
The Finite Element Analysis ◽  
Electric Field Calculation

This article is based on finite element theory and use ANSYS simulation software to establish electric field calculation model of converter transformer for a ±800kV and make electric field calculation and analysis for valve winding. Converter transformer valve winding contour distribution of electric field have completed in the AC, DC and polarity reversal voltage.

scholarly journals Directional selectivity and its use in early visual processing

1981 ◽  
Vol 211 (1183) ◽  
pp. 151-180 ◽  
Keyword(s):  
Visual Processing ◽  
Visual Motion ◽  
Time Derivative ◽  
Receptive Fields ◽  
Specific Model ◽  
Zero Crossings ◽  
Second Stage ◽  
Early Visual Processing ◽  
Local Direction ◽  
Zero Values

The construction of directionally selective units, and their use in the processing of visual motion, are considered. The zero crossings of ∇ 2 G(x, y) ∗ I(x, y) are located, as in Marr & Hildreth (1980). That is, the image is filtered through centre-surround receptive fields, and the zero values in the output are found. In addition, the time derivative ∂[∇ 2 G(x, y) ∗ l(x, y) ]/∂ t is measured at the zero crossings, and serves to constrain the local direction of motion to within 180°. The direction of motion can be determined in a second stage, for example by combining the local constraints. The second part of the paper suggests a specific model of the information processing by the X and Y cells of the retina and lateral geniculate nucleus, and certain classes of cortical simple cells. A number of psychophysical and neurophysiological predictions are derived from the theory.

scholarly journals Open source tools for geographic analysis in transport planning

2021 ◽  
Author(s):  
Robin Lovelace
Keyword(s):  
User Interface ◽  
Open Source ◽  
Citizen Participation ◽  
Simulation Software ◽  
First Century ◽  
Transport Planning ◽  
Geographic Analysis ◽  
Link Type ◽  
Twenty First Century ◽  
Interactive Map

AbstractGeographic analysis has long supported transport plans that are appropriate to local contexts. Many incumbent ‘tools of the trade’ are proprietary and were developed to support growth in motor traffic, limiting their utility for transport planners who have been tasked with twenty-first century objectives such as enabling citizen participation, reducing pollution, and increasing levels of physical activity by getting more people walking and cycling. Geographic techniques—such as route analysis, network editing, localised impact assessment and interactive map visualisation—have great potential to support modern transport planning priorities. The aim of this paper is to explore emerging open source tools for geographic analysis in transport planning, with reference to the literature and a review of open source tools that are already being used. A key finding is that a growing number of options exist, challenging the current landscape of proprietary tools. These can be classified as command-line interface, graphical user interface or web-based user interface tools and by the framework in which they were implemented, with numerous tools released as R, Python and JavaScript packages, and QGIS plugins. The review found a diverse and rapidly evolving ‘ecosystem’ tools, with 25 tools that were designed for geographic analysis to support transport planning outlined in terms of their popularity and functionality based on online documentation. They ranged in size from single-purpose tools such as the QGIS plugin AwaP to sophisticated stand-alone multi-modal traffic simulation software such as MATSim, SUMO and Veins. Building on their ability to re-use the most effective components from other open source projects, developers of open source transport planning tools can avoid ‘reinventing the wheel’ and focus on innovation, the ‘gamified’ A/B Street https://github.com/dabreegster/abstreet/#abstreet simulation software, based on OpenStreetMap, a case in point. The paper, the source code of which can be found at https://github.com/robinlovelace/open-gat, concludes that, although many of the tools reviewed are still evolving and further research is needed to understand their relative strengths and barriers to uptake, open source tools for geographic analysis in transport planning already hold great potential to help generate the strategic visions of change and evidence that is needed by transport planners in the twenty-first century.

Position Selectivity in Scene- and Object-Responsive Occipitotemporal Regions

2007 ◽  
Vol 98 (4) ◽  
pp. 2089-2098 ◽  
Author(s):  
Sean P. MacEvoy ◽  
Russell A. Epstein
Keyword(s):  
Large Scale ◽  
Scene Perception ◽  
Receptive Fields ◽  
Neural Responses ◽  
Retinal Position ◽  
Vertical Meridian ◽  
Retinal Stimulation ◽  
Natural Vision ◽  
Visual Hemifields ◽  
Insight Into

Complex visual scenes preferentially activate several areas of the human brain, including the parahippocampal place area (PPA), the retrosplenial complex (RSC), and the transverse occipital sulcus (TOS). The sensitivity of neurons in these regions to the retinal position of stimuli is unknown, but could provide insight into their roles in scene perception and navigation. To address this issue, we used functional magnetic resonance imaging (fMRI) to measure neural responses evoked by sequences of scenes and objects confined to either the left or right visual hemifields. We also measured the level of adaptation produced when stimuli were either presented first in one hemifield and then repeated in the opposite hemifield or repeated in the same hemifield. Although overall responses in the PPA, RSC, and TOS tended to be higher for contralateral stimuli than for ipsilateral stimuli, all three regions exhibited position-invariant adaptation, insofar as the magnitude of adaptation did not depend on whether stimuli were repeated in the same or opposite hemifields. In contrast, object-selective regions showed significantly greater adaptation when objects were repeated in the same hemifield. These results suggest that neuronal receptive fields (RFs) in scene-selective regions span the vertical meridian, whereas RFs in object-selective regions do not. The PPA, RSC, and TOS may support scene perception and navigation by maintaining stable representations of large-scale features of the visual environment that are insensitive to the shifts in retinal stimulation that occur frequently during natural vision.

Contrast adaptation and contrast gain control

1991 ◽  
Vol 87 (1) ◽  
Author(s):  
L.M. M��tt�nen ◽  
J.J. Koenderink
Keyword(s):  
Gain Control ◽  
Contrast Gain ◽  
Contrast Gain Control ◽  
Contrast Adaptation

Simultaneous Segmentation of Fetal Hearts and Lungs for Medical Ultrasound Images via an Efficient Multi-scale Model Integrated With Attention Mechanism

2021 ◽  
pp. 016173462110425
Author(s):  
Jianing Xi ◽  
Jiangang Chen ◽  
Zhao Wang ◽  
Dean Ta ◽  
Bing Lu ◽  
...  
Keyword(s):  
Congenital Anomaly ◽  
Large Scale ◽  
Automatic Segmentation ◽  
Receptive Fields ◽  
Semantic Segmentation ◽  
Attention Mechanism ◽  
Scale Model ◽  
Ultrasound Images ◽  
Multi Scale ◽  
Task Irrelevant

Large scale early scanning of fetuses via ultrasound imaging is widely used to alleviate the morbidity or mortality caused by congenital anomalies in fetal hearts and lungs. To reduce the intensive cost during manual recognition of organ regions, many automatic segmentation methods have been proposed. However, the existing methods still encounter multi-scale problem at a larger range of receptive fields of organs in images, resolution problem of segmentation mask, and interference problem of task-irrelevant features, obscuring the attainment of accurate segmentations. To achieve semantic segmentation with functions of (1) extracting multi-scale features from images, (2) compensating information of high resolution, and (3) eliminating the task-irrelevant features, we propose a multi-scale model with skip connection framework and attention mechanism integrated. The multi-scale feature extraction modules are incorporated with additive attention gate units for irrelevant feature elimination, through a U-Net framework with skip connections for information compensation. The performance of fetal heart and lung segmentation indicates the superiority of our method over the existing deep learning based approaches. Our method also shows competitive performance stability during the task of semantic segmentations, showing a promising contribution on ultrasound based prognosis of congenital anomaly in the early intervention, and alleviating the negative effects caused by congenital anomaly.

Sign in / Sign up

Export Citation Format

Share Document