Intrinsic Opacity Research Articles

Leveraging saliency priors and explanations for enhanced consistent interpretability

Deep neural networks have emerged as highly effective tools for computer vision systems, showcasing remarkable performance. However, the intrinsic opacity, potential biases, and vulnerability to shortcut learning in these models pose significant concerns regarding their practical application. To tackle these issues, this work employs saliency prior and explanations to enhance the credibility, reliability, and interpretability of neural networks. Specifically, we employ a salient object detection algorithm to extract human-consistent priors from images for data augmentation. The identified saliency priors, along with explanations, serve as supervision signals directing the network’s focus to salient regions within the image. Additionally, contrastive self-supervised learning is incorporated to enable the model to discern the most discriminative concepts. Experimental results confirm the algorithm’s capability to align model explanations with human priors, thereby improving interpretability. Moreover, the proposed approach enhances model performance in data-limited and fine-grained classification scenarios. Importantly, our algorithm is label-independent, allowing for the integration of unlabeled data during training. In practice, this method contributes to improving the reliability and interpretability of intelligent models for downstream tasks. Our code is available here: https://github.com/DLAIResearch/SGC.

62‐4: Invited Paper: Visible‐Blind Infrared Upconversion Devices for Image Sensing

Epitaxial semiconductors dominate infrared sensing applications to date, albeit their intrinsic opacity and brittleness per se limit the possibilities of developing novel technologies. Here, we demonstrate a visible‐blind, all‐organic infrared visualization device that can differentiate the incident infrared power density down to 12 μW cm ‐2 from the intense ambient light illumination of 2000 lux with the help of a hybrid organic‐inorganic metal halide perovskites as a bandpass filter.

When Algorithmic Transparency Failed: Controversies Over Algorithm-Driven Content Curation in the South Korean Digital Environment

Content curation in contemporary digital platforms leverages both algorithmic decision making and human judgment. As much as algorithm has become an integral part of digital configurations, there are growing concerns about the lack of accountability surrounding algorithm-driven digital services. The issue of algorithmic accountability is attributed not only to intrinsic opacity in computational processes but also to the lack of transparency in platform governance. This article discusses two controversial cases surrounding algorithmic transparency in the South Korean digital environment. It first epitomizes the notion of algorithmic transparency as a prerequisite for accountability. Then, it situates the use of algorithms for online content curation in the South Korean digital environment to illustrate how algorithmic transparency is complicated by sociopolitical conditions. Finally, this research offers several suggestions for promoting a more accountable algorithm society.

Confined-Melting-Assisted Synthesis of Bismuth Silicate Glass-Ceramic Nanoparticles: Formation and Optical Thermometry Investigation.

Bismuth-based (nano)materials have been attracting increasing interest due to appealing properties such as high refractive indexes, intrinsic opacity, and structural distortions due to the stereochemistry of 6s2 lone pair electrons of Bi3+. However, the control over specific phases and strategies able to stabilize uniform bismuth-based (nano)materials is still a challenge. In this study, we employed the ability of bismuth to lower the melting point of silica to introduce a new synthetic approach able to confine the growth of bismuth-oxide-based materials into nanostructures. Combining in situ temperature-dependent synchrotron radiation X-ray powder diffraction (XRPD) with high-resolution transmission electron microscopy (HR-TEM) analyses, we demonstrate the evolution of a confined Bi2O3-SiO2 nanosystem from Bi2SiO5 to Bi4Si3O12 through a melting process. The silica shell acts as both a nanoreactor and a silicon source for the stabilization of bismuth silicate glass-ceramic nanocrystals keeping the original spherical shape. The exciton peak of Bi2SiO5 is measured for the first time allowing the estimation of its real energy gap. Moreover, based on a detailed spectroscopic investigation, we discuss the potential and the limitations of Nd3+-activated bismuth silicate systems as ratiometric thermometers. The synthetic strategy introduced here could be further explored to stabilize other bismuth-oxide-based materials, opening the way toward the growth of well-defined glass-ceramic nanoparticles.

MAGICal GRB 190114C: Implications of Cutoff in the Spectrum at sub-GeV Energies

Abstract GRB 190114C is an unusual gamma-ray burst (GRB) due to its detection at sub-TeV energies by MAGIC, seen at redshift z = 0.42. This burst is one of the brightest GRB detected by Fermi. A joint GBM-LAT analysis of the prompt emission reveals the presence of sub-GeV spectral cutoff when the LAT low-energy event (LLE) data is also examined. A similar high-energy cutoff was likewise reported in GRB 160509A and GRB 100724B earlier, as well as a handful of other sources. The cutoff can be explained by the intrinsic opacity due to pair production within the emitting region. GRB 190114C shows a transition from nonthermal to a quasi-thermal-like spectrum and a radiation component that can be attributed to afterglow. Based on spectral analysis, we constrain the site of the prompt emission and Lorentz factor. Knowing that sub-TeV photons are detected in MAGIC, we perceive that the observed spectrum is indeed an overlap from two emission sites, where the emission observed in Fermi is more consistent with prompt emission produced via photospheric dissipation along with a concurrent component from the external shock.

Image generation by GAN and style transfer for agar plate image segmentation

Background and objectives. Deep learning models and specifically Convolutional Neural Networks (CNNs) are becoming the leading approach in many computer vision tasks, including medical image analysis. Nevertheless, the CNN training usually requires large sets of supervised data, which are often difficult and expensive to obtain in the medical field. To address the lack of annotated images, image generation is a promising method, which is becoming increasingly popular in the computer vision community. In this paper, we present a new approach to the semantic segmentation of bacterial colonies in agar plate images, based on deep learning and synthetic image generation, to increase the training set size. Indeed, semantic segmentation of bacterial colony is the basis for infection recognition and bacterial counting in Petri plate analysis. Methods. A convolutional neural network (CNN) is used to separate the bacterial colonies from the background. To face the lack of annotated images, a novel engine is designed — which exploits a generative adversarial network to capture the typical distribution of the bacterial colonies on agar plates — to generate synthetic data. Then, bacterial colony patches are superimposed on existing background images, taking into account both the local appearance of the background and the intrinsic opacity of the bacterial colonies, and a style transfer algorithm is used for further improve visual realism. Results. The proposed deep learning approach has been tested on the only public dataset available with pixel–level annotations for bacterial colony semantic segmentation in agar plates. The role of including synthetic data in the training of a segmentation CNN has been evaluated, showing how comparable performances can be obtained with respect to the use of real images. Qualitative results are also reported for a second public dataset in which the segmentation annotations are not provided. Conclusions. The use of a small set of real data, together with synthetic images, allows obtaining comparable results with respect to using a complete set of real images. Therefore, the proposed synthetic data generator is able to address the scarcity of biomedical data and provides a scalable and cheap alternative to human ground–truth supervision.

The Bright and the Slow—GRBs 100724B and 160509A with High-energy Cutoffs at ≲100 MeV

Abstract We analyze the prompt emission of GRB 100724B and GRB 160509A, two of the brightest gamma-ray bursts (GRBs) observed by Fermi at ≲MeV energies but surprisingly faint at ≳100 MeV energies. Time-resolved spectroscopy reveals a sharp high-energy cutoff at energies E c ∼ 20–60 MeV for GRB 100724B and E c ∼ 80–150 MeV for GRB 160509A. We first characterize phenomenologically the cutoff and its time evolution. We then fit the data to two models where the high-energy cutoff arises from intrinsic opacity to pair production within the source (τ γγ ): (i) a Band spectrum with τ γγ from the internal-shocks-motivated model of Granot et al. (2008) and (ii) the photospheric model of Gill & Thompson (2014). Alternative explanations for the cutoff, such as an intrinsic cutoff in the emitting electron energy distribution, appear to be less natural. Both models provide a good fit to the data with very reasonable physical parameters, providing an estimate of bulk Lorentz factors in the range Γ ∼ 100–400, on the lower end of what is generally observed in Fermi GRBs. Surprisingly, their lower cutoff energies E c compared to other Fermi/LAT GRBs arise not only predominantly from the lower Lorentz factors, but also at a comparable level from differences in variability time, luminosity, and high-energy photon index. Finally, particularly low E c values may prevent detection by Fermi/LAT, thus introducing a bias in the Fermi/LAT GRB sample against GRBs with low Lorentz factors or variability times.

Semitransparent bandages based on chitosan and extracellular matrix for photochemical tissue bonding

BackgroundExtracellular matrices (ECMs) are often used in reconstructive surgery to enhance tissue regeneration and remodeling. Sutures and staples are currently used to fix ECMs to tissue although they can be invasive devices. Other sutureless and less invasive techniques, such as photochemical tissue bonding, cannot be coupled to ECMs because of their intrinsic opacity to light.ResultsWe succeeded in fabricating a biocompatible and adhesive device that is based on ovine forestomach matrix (OFM) and a chitosan adhesive. The natural opacity of the OFM has been overcome by adding the adhesive into the matrix that allows for the light to effectively penetrate through it. The OFM-chitosan device is semitransparent (attenuation length ~ 106 µm) and can be photoactivated by green light to bond to tissue. This device does not require sutures or staples and guarantees a bonding strength of ~ 23 kPa.ConclusionsA new semitransparent and biocompatible bandage has been successfully fabricated and characterized for sutureless tissue bonding.

The extreme blazar AO 0235+164 as seen by extensive ground and space radio observations

Clues to the physical conditions in radio cores of blazars come from measurements of brightness temperatures as well as effects produced by intrinsic opacity. We study the properties of the ultra compact blazar AO 0235+164 with RadioAstron ground-space radio interferometer, multi-frequency VLBA, EVN and single-dish radio observations. We employ visibility modeling and image stacking for deriving structure and kinematics of the source, and use Gaussian process regression to find the relative multi-band time delays of the flares. The multi-frequency core size and time lags support prevailing synchrotron self absorption. The intrinsic brightness temperature of the core derived from ground-based VLBI is close to the equipartition regime value. In the same time, there is evidence for ultra-compact features of the size of less than 10 $\mu$as in the source, which might be responsible for the extreme apparent brightness temperatures of up to $10^{14}$ K as measured by RadioAstron. In 2007--2016 the VLBI components in the source at 43 GHz are found predominantly in two directions, suggesting a bend of the outflow from southern to northern direction. The apparent opening angle of the jet seen in the stacked image at 43 GHz is two times wider than that at 15 GHz, indicating a collimation of the flow within the central 1.5 mas. We estimate the Lorentz factor $\Gamma = 14$, the Doppler factor $\delta=21$, and the viewing angle $\theta = 1.7^\circ$ of the apparent jet base, derive the gradients of magnetic field strength and electron density in the outflow, and the distance between jet apex and the core at each frequency.

Earnings Guidance and Corporate Diversification

This study examines voluntary disclosure differences between single business segment firms and multiple business segment firms. Using management earnings guidance data this paper provides evidence that focused firms are less likely to provide an earnings forecast after controlling for typical determinants of forecast issuance, including various controls for competitive pressures and firm complexity. These results are consistent with higher proprietary costs of disclosure for focused firms potentially stemming from their disclosures being at a finer level of detail. Tests showing that disclosure ranking is not related to excess value are inconsistent with the alternative explanation that diversified firms are more likely to disclose in an attempt to mitigate their intrinsic opacity and therefore reap greater benefits of disclosure.

THE RESPONSE OF METAL-RICH GAS TO X-RAY IRRADIATION FROM A MASSIVE BLACK HOLE AT HIGH REDSHIFT: PROOF OF CONCEPT

Observational studies show that there is a strong link between the formation and evolution of galaxies and the growth of supermassive black holes (SMBH) at their centers. However, the underlying physics of this observed relation is poorly understood. In order to study the effects of X-ray radiation on the surroundings of the black hole, we implement X-ray Dominated Region (XDR) physics into Enzo and use the radiation transport module Moray to calculate the radiative transfer for a polychromatic spectrum. In this work, we investigate the effects of X-ray irradiation, produced by a central massive black hole (MBH) with a mass of M = 5x10^4 M_(solar), on ambient gas with solar and zero metallicity. We find that in the solar metallicity case, due to high opacity of the metals, the energy deposition rate in the central region (< 20 pc) is high and hence the temperatures in the center are on the order of 10^(5-7) K. Moreover, due to the cooling ability and high intrinsic opacity of solar metallicity gas, column densities of 10^(24) cm^(-2) are reached at a radius of 20 pc from the MBH. These column densities are about 3 orders of magnitudes higher than in the zero metallicity case. Furthermore, in the zero metallicity case an X-ray induced H II region is formed already after 5.8 Myr. This causes a significant outflow of gas (~8x10^6 M_(solar) from the central region, with the gas reaching outflow velocities up to ~100 km s^(-1). At later times, ~23 Myr after we insert the MBH, we find that the solar metallicity case also develops an X-ray induced H II region, but delayed by ~17 Myr.

Highly transparent and writable wood all-cellulose hybrid nanostructured paper

Paper, as an inexpensive substrate for flexible electronics and energy devices, has garnered great attention because of its abundance, biodegradability, renewability and sustainability. However, the intrinsic opacity and higher roughness of regular paper greatly restricts further applications. One promising method is to use cellulose nanofibers (CNs) to fabricate nanopaper with a high optical transmittance and excellent smoothness, but there are still some challenges facing nanopaper substrates, such as high-energy consumption to extract nanofibers and the time-consuming process to prepare nanopaper. We design a bilayer hybrid paper using unbeaten wood fibers and CNs with a papermaking technique, which achieves a high optical transmittance and superior smoothness while remaining less expensive than nanopaper and useful as a writable surface. The first transparent paper touchscreen with an excellent anti-glare effect in bright environments is demonstrated using our novel transparent and conductive hybrid paper as the flexible electrode.

Three‐Dimensional Analysis of the Islet Vasculature

The pancreatic islets of Langerhans are highly vascularized structures scattered throughout the pancreas that contain a capillary network 5-10 times denser than that of the exocrine pancreas. A simple method for three-dimensional (3D) analysis of this intricate intraislet vasculature has been difficult because of the intrinsic opacity of the pancreas. We developed a whole-mount imaging technique that allows relatively easy visualization of the islet vasculature. In combination with confocal microscopy and the use of 3D imaging software, we were able to readily reconstruct the 3D architecture of an islet, allowing delineation of the islet volume, length of the intraislet vessels, and the number of vessel branch-points. This technique allows for straightforward 3D image analysis that may help toward understanding islet function.

SIMPLE MODELS OF METAL-LINE ABSORPTION AND EMISSION FROM COOL GAS OUTFLOWS

We analyze the absorption and emission-line profiles produced by a set of simple, cool gas wind models motivated by galactic-scale outflow observations. We implement monte carlo radiative transfer techniques that track the propagation of scattered and fluorescent photons to generate 1D spectra and 2D spectral images. We focus on the MgII 2796,28303 doublet and FeII UV1 multiplet at ~2600A, but the results are applicable to other transitions that trace outflows (e.g. NaI, Lya, SiII). By design, the resonance transitions show blue-shifted absorption but one also predicts strong resonance and fine-structure line-emission at roughly the systemic velocity. This line-emission `fills-in' the absorption reducing the equivalent width by up to 50%, shift the absorption-lin centroid by tens of km/s, and reduce the effective opacity near systemic. Analysis of cool gas outflows that ignores this line-emission may incorrectly infer that the gas is partially covered, measure asignificantly lower peak optical depth, and/or conclude that gas at systemic velocity is absent. Because the FeII lines are connected by optically-thin transitions to fine-structure levels, their profiles more closely reproduce the intrinsic opacity of the wind. Together these results naturally explain the absorption and emission-line characteristics observed for star-forming galaxies at z<1. We also study a scenario promoted to describe the outflows of z~3 Lyman break galaxies and find prfiles inconsistent with the observations due to scattered photon emission. Although line-emission complicates the analysis of absorption-line profiles, the surface brightness profiles offer a unique means of assessing the morphology and size of galactic-scale winds. Furthermore, the kinematics and line-ratios offer powerful diagnostics of outflows, motivating deep, spatially-extended spectroscopic observations.

Three-dimensional optical method for integrated visualization of mouse islet microstructure and vascular network with subcellular-level resolution

Microscopic visualization of islets of Langerhans under normal and diabetic conditions is essential for understanding the pathophysiology of the disease. The intrinsic opacity of pancreata, however, limits optical accessibility for high-resolution light microscopy of islets in situ. Because the standard microtome-based, 2-D tissue analysis confines visualization of the islet architecture at a specific cut plane, 3-D representation of image data is preferable for islet assessment. We applied optical clearing to minimize the random light scattering in the mouse pancreatic tissue. The optical-cleared pancreas allowed penetrative, 3-D microscopic imaging of the islet microstructure and vasculature. Specifically, the islet vasculature was revealed by vessel painting-lipophilic dye labeling of blood vessels-for confocal microscopy. The voxel-based confocal micrographs were digitally processed with projection algorithms for 3-D visualization. Unlike the microtome-based tissue imaging, this optical method for penetrative imaging of mouse islets yielded clear, continuous optical sections for an integrated visualization of the islet microstructure and vasculature with subcellular-level resolution. We thus provide a useful imaging approach to change our conventional planar view of the islet structure into a 3-D panorama for better understanding of the islet physiology.

Microtome-Free 3-Dimensional Confocal Imaging Method for Visualization of Mouse Intestine With Subcellular-Level Resolution

The intrinsic opacity of mouse intestinal tissue prevents its evaluation by high-resolution, in-depth optical microscopy. Instead, intestinal tissue is usually sectioned to expose the interior domains of the mucosa and submucosa for microscopic examination. However, microtome sectioning can cause distortions and artifacts that prevent acquisition of an accurate view of the sample. We therefore attempted to develop a microtome-free 3-dimensional (3D) confocal imaging method for characterization of mouse intestine. We applied an optical-clearing solution, FocusClear, to permeate and reduce the opacity of mouse colon and ileum. Tissues were labeled with fluorescent probes and examined by confocal microscopy with efficient fluorescence excitation and emission in the FocusClear solution. The voxel-based confocal micrographs were processed with Amira software for 3D visualization and analysis. Treatment of tissues with the optical-clearing solution improved photon penetration, resulting in the acquisition of images with subcellular-level resolution across the mucosa, submucosa, and muscle layers. Collectively, the acquired image stacks were processed by projection algorithms for 3D analysis of the spatial relations in villi, crypts, and connective tissues. These imaging technologies allowed for identification of spatiotemporal changes in crypt morphology of colon tissues from mice with dextran sulfate sodium-induced colitis as well as detection of transgenic fluorescent proteins expressed in the colon and ileum. This new optical method for penetrative imaging of mouse intestine does not require tissue sectioning and provides a useful tool for 3D presentation and analysis of diseased and transgenic intestine in an integrated fashion.

On the opacity change required to compensate for the revised solar composition

Context. Recent revisions of the determination of the solar composition have resulted in solar models in marked disagreement with helioseismic inferences. Aims. The effect of the inferred composition change on the models is largely caused by the change in opacity. Thus, we wish to determine an intrinsic opacity change that would compensate for the revision to the composition. Methods. By comparing models computed with the old and revised compositions we determine the required opacity change. Models are computed with the opacity modified in this manner and used as reference in helioseismic inversions to determine the difference between the solar and model sound speed. Results. An opacity increase ranging between around 30 per cent close to the base of the convection zone and a few percent in the solar core leads to a sound-speed profile, for the revised composition, that is essentially indistinguishable from the original solar model. As a function of the logarithm of temperature this is well represented by a simple cubic fit. The physical realism of such a change remains debatable, however.

Mission: impossible (escape from the Lyman limit)

We investigate the intrinsic opacity of high-redshift galaxies to outgoing ionizing photons using high-quality photometry of a sample of 27 spectroscopically identified galaxies of redshift 1.9 > z > 3.5 in the Hubble Deep Field. Our measurement is based on maximum-likelihood fitting of model galaxy spectral energy distributions – including the effects of intrinsic Lyman-limit absorption and random realizations of intervening Lyman-series and Lyman-limit absorption – to photometry of galaxies from space- and ground-based broad-band images. Our method provides several important advantages over the methods used by previous groups, including most importantly that two-dimensional sky subtraction of faint-galaxy images is more robust than one-dimensional sky subtraction of faint-galaxy spectra. We find at the 3σstatistical confidence level that on average no more than 4 per cent of the ionizing photons escape galaxies of redshift 1.9 > z > 3.5. This result is consistent with observations of low- and moderate-redshift galaxies but is in direct contradiction to a recent result based on medium-resolution spectroscopy of high-redshift (z≈ 3) galaxies. Dividing our sample into subsamples according to luminosity, intrinsic ultraviolet colour and redshift, we find no evidence for selection effects that could explain such a discrepancy. Even when all systematic effects are included, the data could not realistically accommodate any escape fraction value larger than ≈15 per cent.

High-resolution confocal imaging and three-dimensional rendering

Intrinsic opacity and inhomeogeniety of most biological tissues have prevented the efficient light penetration and signal detection for high-resolution confocal imaging of thick tissues. Here, we summarize recent technical advances in high-resolution confocal imaging for visualization of cellular structures and gene expression within intact whole-mount thick tissues. First, we introduce features of the FocusClear technology that render biological tissue transparent and thus improve the light penetration and signal detection. Next, a universal fluorescence staining method that labels all nuclei and membranes is described. We then demonstrate the postrecording image processing techniques for 3D visualization. From these images, regions of interest in the whole-mount brain can be segmented and volume rendered. Together, these technical advances in confocal microscopy allow visualization of structures within whole-mount tissues up to 1 mm thick at a resolution similar to that of the observation of single cells in culture. Practical uses and limitations of these techniques are discussed.

The Theatrical Biosphere and Ecologies of Performance

In what would a postmoderntheatrum mundi, or ‘theatre of the world’, consist? In an ironic inversion of the very concept, with the microcosm issuing a unilateral declaration of independence – or of incorporation? Or in a neo-neoplatonic recognition that it is but a cultural construct of an outer world that is itself culturally constructed? In the following article, Baz Kershaw makes connections between the high-imperial Victorian love of glasshouses, which at once created and constrained their ‘theatre of nature’, and the massive 'nineties ecological experiment of ‘Biosphere II’ – ‘a gigantic glass ark the size of an aircraft hangar situated in the Southern Arizona desert’, which embraces all the main types of terrain in the global eco-system. In the Biosphere's ambiguous position between deeply serious scientific experiment and commodified theme park, Kershaw sees an hermetically-sealed system analogous to much contemporary theatre – whose intrinsic opacity is often further blurred by a theorizing no less reductive than that of the obsessive Victorian taxonomists. He offers not answers, but ‘meditations’ on the problem of creating an ecologically meaningful theatre. Baz Kershaw, currently Professor of Drama at the University of Bristol, originally trained and worked as a design engineer. He has had extensive experience as a director and writer in radical theatre, including productions at the Drury Lane Arts Lab and as co-director of Medium Fair, the first mobile rural community arts group, and of the reminiscence theatre company Fair Old Times. He is the author ofThe Politics of Performance: Radical Theatre as Cultural Intervention(Routledge, 1992) andThe Radical in Performance: Between Brecht and Baudrillard(Routledge, 1999), and co-author ofEngineers of the Imagination: the Welfare State Handbook(Methuen, 1990).