Visual Classifications Research Articles

Galaxy Merger Rates up to z ∼ 3 Using a Bayesian Deep Learning Model: A Major-merger Classifier Using IllustrisTNG Simulation Data

Abstract Merging is potentially the dominant process in galaxy formation, yet there is still debate about its history over cosmic time. To address this, we classify major mergers and measure galaxy merger rates up to z ∼ 3 in all five CANDELS fields (UDS, EGS, GOODS-S, GOODS-N, COSMOS) using deep learning convolutional neural networks trained with simulated galaxies from the IllustrisTNG cosmological simulation. The deep learning architecture used is objectively selected by a Bayesian optimization process over the range of possible hyperparameters. We show that our model can achieve 90% accuracy when classifying mergers from the simulation and has the additional feature of separating mergers before the infall of stellar masses from post-mergers. We compare our machine-learning classifications on CANDELS galaxies and compare with visual merger classifications from Kartaltepe et al., and show that they are broadly consistent. We finish by demonstrating that our model is capable of measuring galaxy merger rates, , that are consistent with results found for CANDELS galaxies using close pairs statistics, with . This is the first general agreement between major mergers measured using pairs and structure at z < 3.

Optimizing automatic morphological classification of galaxies with machine learning and deep learning using Dark Energy Survey imaging

ABSTRACT There are several supervised machine learning methods used for the application of automated morphological classification of galaxies; however, there has not yet been a clear comparison of these different methods using imaging data, or an investigation for maximizing their effectiveness. We carry out a comparison between several common machine learning methods for galaxy classification [Convolutional Neural Network (CNN), K-nearest neighbour, logistic regression, Support Vector Machine, Random Forest, and Neural Networks] by using Dark Energy Survey (DES) data combined with visual classifications from the Galaxy Zoo 1 project (GZ1). Our goal is to determine the optimal machine learning methods when using imaging data for galaxy classification. We show that CNN is the most successful method of these ten methods in our study. Using a sample of ∼2800 galaxies with visual classification from GZ1, we reach an accuracy of ∼0.99 for the morphological classification of ellipticals and spirals. The further investigation of the galaxies that have a different ML and visual classification but with high predicted probabilities in our CNN usually reveals the incorrect classification provided by GZ1. We further find the galaxies having a low probability of being either spirals or ellipticals are visually lenticulars (S0), demonstrating that supervised learning is able to rediscover that this class of galaxy is distinct from both ellipticals and spirals. We confirm that ∼2.5 per cent galaxies are misclassified by GZ1 in our study. After correcting these galaxies’ labels, we improve our CNN performance to an average accuracy of over 0.99 (accuracy of 0.994 is our best result).

Agreement Between Visual and Model-Based Classification of Tomato Fruit Ripening

Highlights Human visual classification and predictive models often disagree when only color indices are used. The degree of agreement is improved significantly when predictive models are cultivar-specific. The degree of agreement can be improved when firmness and carotenoid contents are considered. Abstract. Traditionally, the ripening stage of tomato fruit is determined by the observed percentage of red color on the fruit surface based on color charts provided by USDA standards. However, multiple observers can assign different ripening stages to the same tomato fruit due to subjectivity and/or inaccurate evaluations. This practical challenge has not been extensively discussed in the literature, so we assessed the degree of agreement between human visual classification and model-based prediction using physicochemical properties such as color (L*, a*, b*, hue, and chroma), firmness, and carotenoid contents. In our exploratory data analyses, we clearly observed increasing a* and decreasing L*, hue, and firmness with respect to ripening stage, but the rate of change seemed different from cultivar to cultivar. To assess the degree of agreement, cross-validations were used to compare thirty linear regression models with various combinations of the predictors. The cross-validations indicated that predictions from a cultivar-specific model agreed well with human visual classifications. When the cultivar-specific model was considered with the color indices, we achieved up to 95.5% accuracy. When firmness, lycopene, and ß-carotene were added to the model, the accuracy increased to 96.8%. These results suggest the reliability of non-destructive methods for auto-sorting systems. Keywords: ß-carotene, Color index, Firmness, Fruit color, Lycopene, Ripening, Tomato.

Locational Marginal Price Forecasting Using Deep Learning Network Optimized by Mapping-Based Genetic Algorithm

The convolutional neural network (CNN) is commonly used in visual recognitions and classifications. However, CNN can also be applied as a forecaster that can extract features from spatiotemporal data. This paper proposes a 24h ahead electricity price forecasting method, which integrates CNN with an evolutionary algorithm and utilizes spatiotemporal data. The optimal structure of the CNN network for the locational marginal price (LMP) forecasting was obtained using a genetic algorithm (GA). A gene mapping scheme was initially encoded to represent the search space and the process of selection, mutation, and crossover eliminated structures that did not satisfy the validation fitness function and then competitive individuals were generated. The evolution process uses the root mean square error (RMSE) as the validation fitness function, which is optimized by training the created CNN network. The proposed gene mapping scheme can be used to design an optimal CNN structure once the mapping between gene binary bits and parameters/hyperparameters of CNN is given. Day-ahead LMP and demand datasets from Pennsylvania-New Jersey-Maryland (PJM) power market were used to demonstrate the evolutionary capability of the proposed method and the finding of optimal CNN structures. Each studied dataset was grouped into 4 subsets corresponding to various seasonal characteristics (different types of situations in real life). Experimental results revealed that the proposed GA-CNN always yielded a higher forecasting accuracy and lower error rates than other forecasting methods.

Automatic Image Captioning Using Convolution Neural Networks and LSTM

PC vision has turned out to be universal in our general public, with applications in a few fields. Given a lot of pictures, with its inscription, make a prescient model which produces regular, inventive, and intriguing subtitles for the concealed picture. A speedy look at a picture is adequate for a human to call attention to and portray a monstrous measure of insights regarding the visual scene. To rearrange the current issue of producing inscriptions for pictures by making a model which would give exact subtitles to these pictures which can be additionally utilized in other helpful applications and use cases. Be that as it may, this momentous capacity has ended up being a tricky errand for our visual acknowledgment models. Most of the past research in scene acknowledgment has concentrated on naming pictures with a predetermined arrangement of visual classifications and extraordinary advancement has been accom-plished in these undertakings. For a question picture, the past strategies recover pertinent hopeful normal language states by outwardly contrasting the inquiry picture with database pictures. In any case, while shut vocabularies of visual ideas comprise a helpful demonstrating suspicion, they are boundlessly prohibitive when contrasted with the colossal measure of rich depictions that a human can form. These methodologies forced a breaking point on the assortment of inscriptions produced. The model ought to be exempt of suppositions regarding explicit pre decided formats, standards or classes and rather depend on figuring out how to create sentences from the preparation information. The model proposed utilizes Convolution Neural Networks which help to separate highlights of the picture whose subtitle is to be created and afterward by utilizing a probabilistic methodology and Natural Language Processing Techniques reasonable sentences are framed and inscriptions are produced.

CRITICAL SIMULATION: INVESTIGATING THE WORK OF MACHINE VISION IN VISUAL SOCIAL MEDIA CULTURE

Social media are in the midst of an emphatic visual turn (Highfield & Leaver 2016), characterised by the convergence of everyday visual expression with professional creative practice and advertising. The advertising-driven business models of social media platforms increasingly depend on automation. Platforms’ use of machine vision is a key frontier in the algorithmic classification of culture. Machine vision algorithms automatically classify and misclassify faces, expressions, objects, and brand logos in the images users create and share. Images shared by platform users form vast databases used to train these same algorithms. Despite widespread use by social platforms, machine vision is poorly understood and accounted for in the study of everyday visual cultures. In this paper we detail a critical response to the use of automation in visual social media, called critical simulation. We outline a critical simulation framework, the ‘Image Machine’, focussed firstly on Instagram. The Image Machine comprises an Instagram data harvester, and open-access machine vision toolbox that allows digital humanities researchers to interrogate the inner workings of these algorithms and analyse their visual (mis)classifications. In this paper we showcase results from the Image Machine applied to images emanating from a major Australian music festival, Splendour in the Grass. This case examines not only how machine vision classifies and operates on culture, but also how these techniques are being operationalised within the advertising model and promotional culture of platforms like Instagram. We argue that the commercial application of machine vision is interdependent with the participatory culture of platforms like Instagram.

Machine and Deep Learning applied to galaxy morphology - A comparative study

Morphological classification is a key piece of information to define samples of galaxies aiming to study the large-scale structure of the universe. In essence, the challenge is to build up a robust methodology to perform a reliable morphological estimate from galaxy images. Here, we investigate how to substantially improve the galaxy classification within large datasets by mimicking human classification. We combine accurate visual classifications from the Galaxy Zoo project with machine and deep learning methodologies. We propose two distinct approaches for galaxy morphology: one based on non-parametric morphology and traditional machine learning algorithms; and another based on Deep Learning. To measure the input features for the traditional machine learning methodology, we have developed a system called CyMorph, with a novel non-parametric approach to study galaxy morphology. The main datasets employed comes from the Sloan Digital Sky Survey Data Release 7 (SDSS-DR7). We also discuss the class imbalance problem considering three classes. Performance of each model is mainly measured by Overall Accuracy (OA). A spectroscopic validation with astrophysical parameters is also provided for Decision Tree models to assess the quality of our morphological classification. In all of our samples, both Deep and Traditional Machine Learning approaches have over 94.5% OA to classify galaxies in two classes (elliptical and spiral). We compare our classification with state-of-the-art morphological classification from literature. Considering only two classes separation, we achieve 99% of overall accuracy in average when using our deep learning models, and 82% when using three classes. We provide a catalog with 670,560 galaxies containing our best results, including morphological metrics and classification.

VIS3COS

We study the impact of local density and stellar mass on the structure and morphology of approximately 500 quiescent and star-forming galaxies from the VIMOS Spectroscopic Survey of a Superstructure in COSMOS (VIS3COS). We perform bulge-to-disc decomposition of the surface brightness profiles and find ∼41 ± 3% of > 1010 M⊙ galaxies to be best fitted with two components. We complement our analysis with non-parametric morphological measurements and qualitative visual classifications. We find that both galaxy structure and morphology depend on stellar mass and environment for our sample as a whole. We only find an impact of the environment on galaxy size for galaxies more massive than 1011 M⊙. We find higher Sérsic indices (n) and bulge-to-total ratios (B/T) in high-density regions when compared to low-density counterparts at similar stellar masses. We also find that galaxies with higher stellar mass have steeper light profiles (high n, B/T) compared to galaxies with lower stellar mass. Using visual classifications, we find a morphology–density relation at z ∼ 0.84 for galaxies more massive than 1010 M⊙, with elliptical galaxies being dominant at high-density regions and disc galaxies more common in low-density regions. However, when splitting the sample into colour–colour-selected star-forming and quiescent sub-populations, there are no statistically significant differences between low- and high-density regions. We find that quiescent galaxies are smaller, have higher Sérsic indices (for single profiles, around n ∼ 4), and higher bulge-to-total light ratios (for decomposed profiles, around B/T ∼ 0.5) when compared to star-forming counterparts (n ∼ 1 and B/T ∼ 0.3, for single and double profiles, respectively). We confirm these trends with non-parametric quantities, finding quiescent galaxies to be smoother (lower asymmetry, lower M20) and to have most of their light over smaller areas (higher concentration and Gini coefficient) than star-forming galaxies. Overall, we find a stronger dependence of structure and morphology on stellar mass than on local density and these relations are strongly correlated with the quenching fraction. The change in average structure or morphology corresponds to a change in the relative fractions of blue disc-like galaxies and red elliptical galaxies with stellar mass and environment. We hypothesise that the processes responsible for the quenching of star formation must also affect the galaxy morphology on similar timescales.

Classification of Broad Absorption Line Quasars with a Convolutional Neural Network

Abstract Quasars that exhibit blueshifted, broad absorption lines (BAL QSOs) are an important probe of black hole feedback on galaxy evolution. Yet the presence of BALs is also a complication for large spectroscopic surveys that use quasars as cosmological probes because the BAL features can affect redshift measurements and contaminate information about the matter distribution in the Lyα forest. We present a new BAL QSO catalog for quasars in the Sloan Digital Sky Survey (SDSS) Data Release 14 (DR14). As the SDSS DR14 quasar catalog has over 500,000 quasars, we have developed an automated BAL classifier with a Convolutional Neural Network (CNN). We trained our CNN classifier on the C iv λ 1549 region of a sample of quasars with reliable human classifications, and compared the results to both a dedicated test sample and visual classifications from the earlier SDSS DR12 quasar catalog. Our CNN classifier correctly classifies over 98% of the BAL quasars in the DR12 catalog, which demonstrates comparable reliability to human classification. The disagreements are generally for quasars with lower signal-to-noise ratio spectra and/or weaker BAL features. Our new catalog includes the probability that each quasar is a BAL, the strength, blueshifts and velocity widths of the troughs, and similar information for any Si iv λ 1398 BAL troughs that may be present. We find significant BAL features in 16.8% of all quasars with 1.57 < z < 5.56 in the SDSS DR14 quasar catalog.

Su1744 DIAGNOSTIC ACCURACY AND LEARNING CURVE IN THE CHARACTERIZATION OF COLORECTAL POLYPS FOR FIRST YEAR GASTROINTESTINAL ENDOSCOPY TRAINEES

Identification and removal of colorectal adenomas during colonoscopy has been associated with risk reduction in colorectal cancer. Multiple studies have shown the utility of endoscopic characterization of polyps in predicting a histopathologic diagnosis. The use of narrow band imaging and visual classifications such as NBI International Colorectal Endoscopic (NICE) and Japan NBI Expert Team classification (JNET) have been associated with a high diagnostic accuracy in the distinction between neoplastic and non-neoplastic lesions with a short learning curve.

Studying the physical properties of tidal features – I. Extracting morphological substructure in CANDELS observations and VELA simulations

Abstract The role of major mergers in galaxy evolution remains a key open question. Existing empirical merger identification methods use non-parametric and subjective visual classifications that can pose systematic challenges to constraining merger histories. As a first step towards overcoming these challenges, we develop and share publicly a new python-based software tool that identifies and extracts the flux-wise and area-wise significant contiguous regions from the model-subtracted residual images produced by popular parametric light-profile fitting tools (e.g. galfit). Using Hubble Space Telescope (HST) H-band single-Sérsic residual images of 17 CANDELS galaxies, we demonstrate the tools ability to measure the surface brightness and improve the qualitative identification of a variety of common residual features (disc structures, spiral substructures, plausible tidal features, and strong gravitational arcs). We test our method on synthetic HST observations of a z ∼ 1.5 major merger from the VELA hydrodynamic simulations. We extract H-band residual features corresponding to the birth, growth, and fading of tidal features during different stages and viewing orientations at CANDELS depths and resolution. We find that the extracted features at shallow depths have noisy visual appearance and are susceptible to viewing angle effects. For a VELA z ∼ 3 major merger, we find that James Webb Space Telescope NIRCam observations can probe high-redshift tidal features with considerable advantage over existing HST capabilities. Further quantitative analysis of plausible tidal features extracted with our new software hold promise for the robust identification of hallmark merger signatures and corresponding improvements to merger rate constraints.

Galaxy Merger Fractions in Two Clusters at Using the Hubble Space Telescope

Abstract We measure the fraction of galaxy–galaxy mergers in two clusters at z ∼ 2 using imaging and grism observations from the Hubble Space Telescope. The two galaxy cluster candidates were originally identified as overdensities of objects using deep mid-infrared imaging and observations from the Spitzer Space Telescope, and were subsequently followed up with HST/WFC3 imaging and grism observations. We identify galaxy–galaxy merger candidates using high-resolution imaging with the WFC3 in the F105W, F125W, and F160W bands. Coarse redshifts for the same objects are obtained with grism observations in G102 for the z ∼ 1.6 cluster (IRC0222A) and G141 for the z ∼ 2 cluster (IRC0222B). Using visual classifications as well as a variety of selection techniques, we measure merger fractions of in IRC0222A and in IRC0222B. In comparison, we measure a merger fraction of for field galaxies at z ∼ 2. Our study indicates that the galaxy–galaxy merger fraction in clusters at z ∼ 2 is enhanced compared to the field population, but note that more cluster measurements at this epoch are needed to confirm our findings.

Evolution of Silcrete Heat Treatment in Australia—a Regional Pattern on the South-East Coast and Its Evolution over the Last 25 ka

We document long-term evolution in the rate of heat treatment in Eastern Australia and explore them as a technological response to dynamic industrial and social contexts that developed in the last 25 millennia. We employed methods previously used in Africa but novel in Australia to infer long-term directional changes in the relative frequency of silcrete artefacts that were heat-treated. Our methods involved independent and cross-verifying tests of the presence or absence of heat treatment, employing visual classifications and surface roughness measures. These methods revealed a coherent series of increases over time in the landscapes around Sydney, so that Late Holocene assemblages displayed higher rates of heat treatment than terminal Pleistocene or Early Holocene ones. We hypothesise that the directional trend towards greater frequencies of heat-treated artefacts on Australia’s eastern seaboard is explicable in terms of the context of technological shifts towards microlith production superimposed onto an even longer term process of lithic resource depletion, perhaps compounded with the development of political barriers to the redistribution of knappable stone.

Transfer learning for galaxy morphology from one survey to another

Deep Learning (DL) algorithms for morphological classification of galaxies have proven very successful, mimicking (or even improving) visual classifications. However, these algorithms rely on large training samples of labelled galaxies (typically thousands of them). A key question for using DL classifications in future Big Data surveys is how much of the knowledge acquired from an existing survey can be exported to a new dataset, i.e. if the features learned by the machines are meaningful for different data. We test the performance of DL models, trained with Sloan Digital Sky Survey (SDSS) data, on Dark Energy survey (DES) using images for a sample of $\sim$5000 galaxies with a similar redshift distribution to SDSS. Applying the models directly to DES data provides a reasonable global accuracy ($\sim$ 90%), but small completeness and purity values. A fast domain adaptation step, consisting in a further training with a small DES sample of galaxies ($\sim$500-300), is enough for obtaining an accuracy > 95% and a significant improvement in the completeness and purity values. This demonstrates that, once trained with a particular dataset, machines can quickly adapt to new instrument characteristics (e.g., PSF, seeing, depth), reducing by almost one order of magnitude the necessary training sample for morphological classification. Redshift evolution effects or significant depth differences are not taken into account in this study.

Tidal Interactions and Mergers in Intermediate-redshift EDisCS Clusters

Abstract We study the fraction of tidal interactions and mergers (TIMs) with well-identified observability timescales (f TIM) in group, cluster, and accompanying field galaxies and its dependence on redshift (z), cluster velocity dispersion (σ), and environment analyzing Hubble Space Telescope/ACS images and catalogs from the ESO Distant Cluster Survey. Our sample consists of 11 clusters, seven groups, and accompanying field galaxies at 0.4 ≤ z ≤ 0.8. We derive f TIM using both a visual classification of galaxy morphologies and an automated method, the G − M 20 method. We calibrate this method using the visual classifications that were performed on a subset of our sample. We find marginal evidence for a trend between f TIM and z, in that higher z values correspond to higher f TIM. However, we also cannot rule out the null hypothesis of no correlation at higher than 68% confidence. No trend is present between f TIM and σ. We find that f TIM shows suggestive peaks in groups, and tentatively in clusters at R > 0.5 × R 200, implying that f TIM gets boosted in these intermediate-density environments. However, our analysis of the local densities of our cluster sample does not reveal a trend between f TIM and density, except for a potential enhancement at the very highest densities. We also perform an analysis of projected radius–velocity phase space for our cluster members. Our results reveal that TIM and undisturbed galaxies only have a 6% probability of having been drawn from the same parent population in their velocity distribution and 37% in radii, in agreement with the modest differences obtained in f TIM at the clusters.

The Fornax Deep Survey with the VST

Context. The Fornax Deep Survey (FDS), an imaging survey in the u′, g′, r′, and i′-bands, has a supreme resolution and image depth compared to the previous spatially complete Fornax Cluster Catalog (FCC). Our new data allows us to study the galaxies down to r′-band magnitude mr′ ≈ 21 mag (Mr′ ≈ −10.5 mag), which opens a new parameter regime to investigate the evolution of dwarf galaxies in the cluster environment. After the Virgo cluster, Fornax is the second nearest galaxy cluster to us, and with its different mass and evolutionary state, it provides a valuable comparison that makes it possible to understand the various evolutionary effects on galaxies and galaxy clusters. These data provide an important legacy dataset to study the Fornax cluster. Aims. We aim to present the Fornax Deep Survey (FDS) dwarf galaxy catalog, focusing on explaining the data reduction and calibrations, assessing the quality of the data, and describing the methods used for defining the cluster memberships and first order morphological classifications for the catalog objects. We also describe the main scientific questions that will be addressed based on the catalog. This catalog will also be invaluable for future follow-up studies of the Fornax cluster dwarf galaxies. Methods. As a first step we used the SExtractor fine-tuned for dwarf galaxy detection, to find galaxies from the FDS data, covering a 26 deg2 area of the main cluster up to its virial radius, and the area around the Fornax A substructure. We made 2D-decompositions of the identified galaxies using GALFIT, measure the aperture colors, and the basic morphological parameters like concentration and residual flux fraction. We used color–magnitude, luminosity–radius and luminosity–concentration relations to separate the cluster galaxies from the background galaxies. We then divided the cluster galaxies into early- and late-type galaxies according to their morphology and gave first order morphological classifications using a combination of visual and parametric classifications. Results. Our final catalog includes 14 095 galaxies. We classify 590 galaxies as being likely Fornax cluster galaxies, of which 564 are dwarfs (Mr′ > −18.5 mag) consisting our Fornax dwarf catalog. Of the cluster dwarfs we classify 470 as early-types, and 94 as late-type galaxies. Our final catalog reaches its 50% completeness limit at magnitude Mr′ = −10.5 mag and surface brightness μ¯e,r′ = 26 mag arcsec−2, which is approximately three magnitudes deeper than the FCC. Based on previous works and comparison with a spectroscopically confirmed subsample, we estimate that our final Fornax dwarf galaxy catalog has ≲10% contamination from the background objects.

Identification of low surface brightness tidal features in galaxies using convolutional neural networks

Faint tidal features around galaxies record their merger and interaction histories over cosmic time. Due to their low surface brightnesses and complex morphologies, existing automated methods struggle to detect such features and most work to date has heavily relied on visual inspection. This presents a major obstacle to quantitative study of tidal debris features in large statistical samples, and hence the ability to be able to use these features to advance understanding of the galaxy population as a whole. This paper uses convolutional neural networks (CNNs) with dropout and augmentation to identify galaxies in the CFHTLS-Wide Survey that have faint tidal features. Evaluating the performance of the CNNs against previously-published expert visual classifications, we find that our method achieves high (76%) completeness and low (20%) contamination, and also performs considerably better than other automated methods recently applied in the literature. We argue that CNNs offer a promising approach to effective automatic identification of low surface brightness tidal debris features in and around galaxies. When applied to forthcoming deep wide-field imaging surveys (e.g. LSST, Euclid), CNNs have the potential to provide a several order-of-magnitude increase in the sample size of morphologically-perturbed galaxies and thereby facilitate a much-anticipated revolution in terms of quantitative low surface brightness science.

BOLD mapping of human epileptic spikes recorded during simultaneous intracranial EEG-fMRI: The impact of automated spike classification

ObjectivesSimultaneous intracranial EEG and functional MRI (icEEG-fMRI) can be used to map the haemodynamic (BOLD) changes associated with the generation of IEDs. Unlike scalp EEG-fMRI, in most patients who undergo icEEG-fMRI, IEDs recorded intracranially are numerous and show variability in terms of field amplitude and morphology. Therefore, visual marking can be highly subjective and time consuming. In this study, we applied an automated spike classification algorithm, Wave_clus (WC), to IEDs marked visually on icEEG data acquired during simultaneous fMRI acquisition. The motivation of this work is to determine whether using a potentially more consistent and unbiased automated approach can produce more biologically meaningful BOLD patterns compared to the BOLD patterns obtained based on the conventional, visual classification. MethodsWe analysed simultaneous icEEG-fMRI data from eight patients with severe drug resistant epilepsy, and who subsequently underwent resective surgery that resulted in a good outcome: confirmed epileptogenic zone (EZ). For each patient two fMRI analyses were performed: one based on the conventional visual IED classification and the other based on the automated classification. We used the concordance of the IED-related BOLD maps with the confirmed EZ as an indication of their biological meaning, which we compared for the automated and visual classifications for all IED originating in the EZ. ResultsAcross the group, the visual and automated classifications resulted in 32 and 24 EZ IED classes respectively, for which 75% vs 83% of the corresponding BOLD maps were concordant. At the single-subject level, the BOLD maps for the automated approach had greater concordance in four patients, and less concordance in one patient, compared to those obtained using the conventional visual classification, and equal concordance for three remaining patients. These differences did not reach statistical significance. ConclusionWe found automated IED classification on icEEG data recorded during fMRI to be feasible and to result in IED-related BOLD maps that may contain similar or greater biological meaning compared to the conventional approach in the majority of the cases studied. We anticipate that this approach will help to gain significant new insights into the brain networks associated with IEDs and in relation to postsurgical outcome.

What can a low-cost fish-finder tell us about the seabed?

Low-cost non-scientific echosounders are commonly used for safety and other non-scientific purposes. These echosounders routinely record data which are rarely processed for later use, but which could provide interesting information about the seabed. In this work, we report on the use of a low-cost echosounder for bottom classification in five coastal rocky reefs of northern Argentine Patagonia. Underwater video transects were acquired over the same area as the acoustic data, in order to compare acoustic and visual classifications at or near the points where acoustic and video transects intersected (crossing sites). The reliability of these validation points decreases with the distance between the acoustic data points and the video transects as well as with the habitat heterogeneity as observed in the video. The shape of the depth-corrected acoustic echoes allowed easy discrimination between hard (or rocky) and soft (or muddy) bottoms by using a linear function as threshold optimized from a training set of crossing sites. The classification accuracy obtained using only validation points classified as most reliable was 96% (Cohen's kappa coefficient κ = 0.88, indicating nearly perfect agreement). These results show that this low-cost methodology offers a suitable alternative to using scientific echosounders for mapping rocky-reef areas.

Multiview Label Sharing for Visual Representations and Classifications

Different views represent different aspects of images. It is more effective to combine them for visual classifications. This paper proposes a novel multiview label sharing method to combine the discriminative power of different views for classifications. Especially, we linearly transfer different views into a shared space for representations. The inter-view similarities are kept in the shared space for each view. We also ensure the intra-view similarities of the same class between different views are preserved in the shared space. We jointly learn the classifiers and transformation matrices by minimizing the summed classification loss along with the inter-view and intra-view similarity constraints. In this paper, the inter-view constraints refer to the similarities between images of the corresponding view, whereas the intra-view constraints refer to the similarities between different views of images with the same semantics. Experimental results and analysis on several public datasets show the effectiveness of the proposed multiview label sharing method for visual classifications.