Search

In real-world scenarios we often need to perform multiple tasks simultaneously. Multi-Task Learning (MTL) is an adequate method to do so, but usually requires datasets labeled for all tasks. We propose a method that can leverage datasets labeled for only some of the tasks in the MTL framework. Our work, Knowledge Assembly (KA), learns multiple tasks from disjoint datasets by leveraging the unlabeled data in a semi-supervised manner, using model augmentation for pseudo-supervision. Whilst KA can be implemented on any existing MTL networks, we test our method on jointly learning person re-identification (reID) and pedestrian attribute recognition (PAR). We surpass the single task fully-supervised performance by $4.2\%$ points for reID and $0.9\%$ points for PAR., Comment: Accepted at CVPRW'23

Significance: Multiple eye diseases such as age-related macular degeneration, diabetic retinopathy, and retinitis pigmentosa can cause photoreceptor dysfunction. Rod photoreceptors are known to be more vulnerable than cone photoreceptors. Therefore, functional assessment of rod photoreceptors is important for early detection of eye diseases. Aim: This study is to demonstrate the feasibility of using intrinsic optical signal (IOS) optoretinography (ORG) for objective detection of dark adaptation (DA) abnormality due to rod photoreceptor degeneration. Approach: Functional optical coherence tomography (OCT) was employed for IOS ORG of wild-type (WT) and retinal degeneration 10 (rd10) mice. Six WT C57BL/6J and eight rd10 B6.CXB1-Pde6brd10/J mice at postnatal day 14 (P14) were used for this study. Dynamic OCT analysis of retinal thickness and brightness changes, corresponding to light to dark transition, was implemented Results: Comparative measurement of the retina under light and dark conditions revealed significant IOS changes within the outer retina. The thickness between external limiting membrane (ELM) and retinal pigment epithelium (RPE) reduced, and the OCT brightness of inner segment ellipsoid zone (EZ) decreased during DA, compared to light adaptation (LA). Relative EZ intensity change was observed to decrease larger in rd10, compared to WT. The relative intensity of the hyporeflective band between ELM and RPE also showed significant decrease in rd10 retinas during DA. Conclusions: DA-IOS abnormalities were observed in rd10, compared to WT at P14. The ORG measurement of DA-IOS kinetics promises valuable information for noninvasive assessment of photoreceptor function.

Improving multi-view aggregation is integral for multi-view pedestrian detection, which aims to obtain a bird's-eye-view pedestrian occupancy map from images captured through a set of calibrated cameras. Inspired by the success of attention modules for deep neural networks, we first propose a Homography Attention Module (HAM) which is shown to boost the performance of existing end-to-end multiview detection approaches by utilizing a novel channel gate and spatial gate. Additionally, we propose Booster-SHOT, an end-to-end convolutional approach to multiview pedestrian detection incorporating our proposed HAM as well as elements from previous approaches such as view-coherent augmentation or stacked homography transformations. Booster-SHOT achieves 92.9% and 94.2% for MODA on Wildtrack and MultiviewX respectively, outperforming the state-of-the-art by 1.4% on Wildtrack and 0.5% on MultiviewX, achieving state-of-the-art performance overall for standard evaluation metrics used in multi-view pedestrian detection., Comment: Arxiv preprint

Deep learning models are increasingly deployed in real-world applications. These models are often deployed on the server-side and receive user data in an information-rich representation to solve a specific task, such as image classification. Since images can contain sensitive information, which users might not be willing to share, privacy protection becomes increasingly important. Adversarial Representation Learning (ARL) is a common approach to train an encoder that runs on the client-side and obfuscates an image. It is assumed, that the obfuscated image can safely be transmitted and used for the task on the server without privacy concerns. However, in this work, we find that training a reconstruction attacker can successfully recover the original image of existing ARL methods. To this end, we introduce a novel ARL method enhanced through low-pass filtering, limiting the available information amount to be encoded in the frequency domain. Our experimental results reveal that our approach withstands reconstruction attacks while outperforming previous state-of-the-art methods regarding the privacy-utility trade-off. We further conduct a user study to qualitatively assess our defense of the reconstruction attack., Comment: The IJCAI-ECAI-22 Workshop on Artificial Intelligence Safety (AISafety 2022)

Niobium is commonly used for superconducting quantum systems as readout resonators, capacitors, and interconnects. The coherence time of the superconducting qubits is mainly limited by microwave dissipation attributed to two-level system defects at interfaces, such as the Nb/Si and Nb/air interface. One way to improve the Nb/air interface quality is by thermal annealing, as shown by extensive studies in 3D superconducting radio frequency (SRF) cavities. However, it is unclear how the microstructure and chemistry of the interface structures change during heat treatment. To address this knowledge gap, we comprehensively characterized Nb films deposited on Si wafers by physical vapor deposition, including (1) an Nb film from a transmon and (2) an Nb film without any patterning step, using an aberration-corrected transmission electron microscope. Both Nb films exhibit columnar growth with strong [110] textures. There is a double layer between the Nb film and Si substrate, which are amorphous niobium silicides with different Nb and Si concentrations. After in-situ heating of the heterostructure at 360{\deg}C inside the microscope, the composition of the double layers at the Nb-Si interface remains almost the same despite different thickness changes. The initial amorphous niobium oxide layer on Nb surface decomposes into face-centered cubic Nb nanograins in the amorphous Nb-O matrix upon heating.

Intrinsic optical signal (IOS) imaging of the retina, also termed as optoretinography (ORG), promises a noninvasive method for objective assessment of retinal function. By providing unparalleled capability to differentiate individual layers of the retina, functional optical coherence tomography (OCT) has been actively investigated for intrinsic signal ORG measurements. However, clinical deployment of functional OCT for quantitative ORG is still challenging due to the lack of a standardized imaging protocol and the complication of IOS sources and mechanisms. This article aims to summarize recent developments of functional OCT for ORG measurement, OCT intensity- and phase-based IOS processing. Technical challenges and perspectives of quantitative IOS analysis and ORG interpretations are discussed.

Delayed dark adaptation due to impaired rod photoreceptor homeostasis has been reported as the earliest symptom of eye diseases such as age-related macular degeneration, diabetic retinopathy, and retinitis pigmentosa. Objective measurement of dark adaptation can facilitate early diagnosis to enable prompt intervention to prevent vision losses. However, there is a lack of noninvasive methods capable of spatiotemporal monitoring of photoreceptor changes during dark adaptation. Here we demonstrate functional optical coherence tomography (OCT) for in vivo intrinsic signal optoretinography (ORG) of dark adaptation kinetics in the C57BL/6J mouse retina. Functional OCT revealed a shortening of the outer retina, a morphological change in the cone and rod photoreceptor interdigitation zone, and a reduction in intrinsic signal amplitude at the photoreceptor inner segment ellipsoid. A strong positive correlation between morphophysiological activities was also confirmed. Functional OCT of dark adaptation kinetics promises a method for rapid ORG assessment of physiological integrity of retinal photoreceptors., Comment: 18 pages, 7 figures

This study is to characterize reflectance profiles of retinal blood vessels in optical coherence tomography (OCT), and to validate these vascular features to guide artery-vein classification in OCT angiography (OCTA) of human retina. Depth-resolved OCT reveals unique features of retinal arteries and veins. Retinal arteries show hyper-reflective boundaries at both upper (inner side towards the vitreous) and lower (outer side towards the choroid) walls. In contrary, retinal veins reveal hyper-reflectivity at the upper boundary only. Uniform lumen intensity was observed in both small and large arteries. However, the vein lumen intensity was dependent on the vessel size. Small veins exhibit a hyper-reflective zone at the bottom half of the lumen, while large veins show a hypo-reflective zone at the bottom half of the lumen, Comment: 11 pages, 4 figures

External magnetic field, temperature, and spin-polarized current are usually employed to create and control nanoscale vortex-like spin configurations such as magnetic skyrmions. Although these methods have proven successful, they are not energy-efficient due to high power consumption and dissipation. Coupling between magnetic properties and mechanical deformation, the magnetoelastic (MEL) effect, offers a novel approach to energy-efficient control of magnetism at the nanoscale. It is of great interest in the context of ever-decreasing length scales of electronic and spintronic devices. Therefore, it is desirable to establish a comprehensive framework capable of predicting the effects of mechanical stress and enabling deterministic control of magnetic textures and skyrmions. In this work, using an advanced scheme of multiscale simulations and Lorentz transmission electron microscopy measurements we demonstrate deterministic control of topological magnetic textures and skyrmion creation in thin films and racetracks of chiral magnets. Our investigation considers not only uniaxial but also biaxial stress, which is ubiquitous in thin-film devices. The biaxial stress, rather than the uniaxial one, was shown to be more efficient to create or annihilate skyrmions when the MEL coefficient and strain have the same or opposite signs, respectively. It was also demonstrated to be a viable way to stabilize skyrmions and to control their current-induced motion in racetrack memory. Our results open prospects for deployment of mechanical stress to create novel topological spin textures, including merons, and in control and optimization of skyrmion-based devices., Comment: 12 pages, 7 figures

Magnetic Weyl semimetals are expected to have extraordinary physical properties such as a chiral anomaly and large anomalous Hall effects that may be useful for future, potential, spintronics applications. However, in most known host materials, multiple pairs of Weyl points prevent a clear manifestation of the intrinsic topological effects. Our recent density functional theory (DFT) calculations study suggest that EuCd$_{2}$As$_{2}$ can host Dirac fermions in an antiferromagnetically (AFM) ordered state or a single pair of Weyl fermions in a ferromagnetically (FM) ordered state. Unfortunately, previously synthesized crystals ordered antiferromagnetically with $T_{\textrm{N}}$\,$\simeq$\,9.5\,K. Here, we report the successful synthesis of single crystals of EuCd$_{2}$As$_{2}$ that order ferromagnetically (FM) or antiferromagnetically (AFM) depending on the level of band filling, thus allowing for the use of magnetism to tune the topological properties within the same host. We explored their physical properties via magnetization, electrical transport, heat capacity, and angle resolved photoemission spectroscopy (ARPES) measurements and conclude that EuCd$_{2}$As$_{2}$ is an excellent, tunable, system for exploring the interplay of magnetic ordering and topology.

Complex polar structures of incommensurate modulations (ICMs) are revealed in chemically modified PbZrO$_3$ perovskite antiferroelectrics using advanced transmission electron microscopy techniques. The Pb-cation displacements, previously assumed to arrange in a fully-compensated antiparallel fashion, are found to be either antiparallel but with different magnitudes, or in a nearly orthogonal arrangement in adjacent stripes in the ICMs. Ab initio calculations corroborate the low-energy state of these arrangements. Our discovery corrects the atomic understanding of ICMs in PbZrO$_3$-based perovskite antiferroelectrics.

As there are increasing needs of sharing data for machine learning, there is growing attention for the owners of the data to claim the ownership. Visible watermarking has been an effective way to claim the ownership of visual data, yet the visibly watermarked images are not regarded as a primary source for learning visual recognition models due to the lost visual information by in the watermark and the possibility of an attack to remove the watermarks. To make the watermarked images better suited for machine learning with less risk of removal, we propose DeepStamp, a watermarking framework that, given a watermarking image and a trained network for image classification, learns to synthesize a watermarked image that are human-perceptible, robust to removals, and able to be used as training images for classification with minimal accuracy loss. To achieve the goal, we employ the generative multi-adversarial network (GMAN). In experiments with CIFAR10, we show that the DeepStamp learn to transform a watermark to be embedded in each image and the watermarked images can be used to train networks., Comment: Presented as a poster at the NDSS'19

High-performance visual recognition systems generally require a large collection of labeled images to train. The expensive data curation can be an obstacle for improving recognition performance. Sharing more data allows training for better models. But personal and private information in the data prevent such sharing. To promote sharing visual data for learning a recognition model, we propose to obfuscate the images so that humans are not able to recognize their detailed contents, while machines can still utilize them to train new models. We validate our approach by comprehensive experiments on three challenging visual recognition tasks; image classification, attribute classification, and facial landmark detection on several datasets including SVHN, CIFAR10, Pascal VOC 2012, CelebA, and MTFL. Our method successfully obfuscates the images from humans recognition, but a machine model trained with them performs within about 1% margin (up to 0.48%) of the performance of a model trained with the original, non-obfuscated data., Comment: The logical flow and the experimental validations have to be significantly revised

To reduce material and processing costs of commercial permanent magnets and to attempt to fill the empty niche of energy products, 10 - 20 MGOe, between low-flux (ferrites, alnico) and high-flux (Nd2Fe14B- and SmCo5-type) magnets, we report synthesis, structure, magnetic properties and modeling of Ta, Cu and Fe substituted CeCo5. Using a self-flux technique, we grew single crystals of I - Ce15.1Ta1.0Co74.4Cu9.5, II - Ce16.3Ta0.6Co68.9Cu14.2, III - Ce15.7Ta0.6Co67.8Cu15.9, IV - Ce16.3Ta0.3Co61.7Cu21.7 and V - Ce14.3Ta1.0Co62.0Fe12.3Cu10.4. X-ray diffraction analysis (XRD) showed that these materials retain a CaCu5 substructure and incorporate small amounts of Ta in the form of \dumb-bells", filling the 2e crystallographic sites within the 1D hexagonal channel with the 1a Ce site, whereas Co, Cu and Fe are statistically distributed among the 2c and 3g crystallographic sites. Scanning electron microscopy, energy dispersive X-ray spectroscopy (SEM-EDS) and scanning transmission electron microscopy (STEM) examinations provided strong evidence of the single-phase nature of the as-grown crystals, even though they readily exhibited significant magnetic coercivitie of ~1.6 - ~1.8 kOe caused by Co-enriched, nano-sized, structural defects and faults that can serve as pinning sites. Formation of the "composite crystal" during the heat treatment creates a 3D array of extended defects within a primarily single grain single crystal, which greatly improves its magnetic characteristics. Possible causes of the formation of the "composite crystal" may be associated with Ta atoms leaving matrix interstices at lower temperatures and/or matrix degradation induced by decreased miscibility at lower temperatures. Fe strongly improves both the Curie temperature and magnetization of the system resulting in (BH)max:~13 MGOe at room temperature., Comment: 17 pages, 18 figures

Single crystals of AlFe$_{2}$B$_{2}$ have been grown using the self flux growth method and then measured the structural properties, temperature and field dependent magnetization, and temperature dependent electrical resistivity at ambient as well as high pressure. The Curie temperature of AlFe$_{2}$B$_{2}$ is determined to be $274$~K. The measured saturation magnetization and the effective moment for paramagnetic Fe-ion indicate the itinerant nature of the magnetism with a Rhode-Wohlfarth ratio $ \frac{M_{C}}{M_{sat}}\approx 1.14$. Temperature dependent resistivity measurements under hydrostatic pressure shows that transition temperature \textit{T$_C$} is suppressed down to 255 K for $p = 2.24$~GPa pressure with a suppression rate of $\sim -8.9$~K/GPa. The anisotropy fields and magnetocrystalline anisotropy constants are in reasonable agreement with density functional theory calculations., Comment: 12 pages, 12 figures

Quantum states of quasiparticles in solids are dictated by symmetry. Thus, a discovery of unconventional symmetry can provide a new opportunity to reach a novel quantum state. Recently, Dirac and Weyl electrons have been observed in crystals with discrete translational symmetry. Here we experimentally demonstrate Dirac electrons in a two-dimensional quasicrystal without translational symmetry. A dodecagonal quasicrystal was realized by epitaxial growth of twisted bilayer graphene rotated exactly 30 degree. The graphene quasicrystal was grown up to a millimeter scale on SiC(0001) surface while maintaining the single rotation angle over an entire sample and was successfully isolated from a substrate, demonstrating its structural and chemical stability under ambient conditions. Multiple Dirac cone replicated with the 12-fold rotational symmetry were observed in angle resolved photoemission spectra, showing its unique electronic structures with anomalous strong interlayer coupling with quasi-periodicity. Our study provides a new way to explore physical properties of relativistic fermions with controllable quasicrystalline orders., Comment: 20 pages, 4 figures

We propose to learn a curriculum or a syllabus for supervised learning and deep reinforcement learning with deep neural networks by an attachable deep neural network, called ScreenerNet. Specifically, we learn a weight for each sample by jointly training the ScreenerNet and the main network in an end-to-end self-paced fashion. The ScreenerNet neither has sampling bias nor requires to remember the past learning history. We show the networks augmented with the ScreenerNet achieve early convergence with better accuracy than the state-of-the-art curricular learning methods in extensive experiments using three popular vision datasets such as MNIST, CIFAR10 and Pascal VOC2012, and a Cart-pole task using Deep Q-learning. Moreover, the ScreenerNet can extend other curriculum learning methods such as Prioritized Experience Replay (PER) for further accuracy improvement., Comment: curricular learning, deep learning, deep q-learning

Vertical and lateral heterogeneous structures of two-dimensional (2D) materials have paved the way for pioneering studies on the physics and applications of 2D materials. A hybridized hexagonal boron nitride (h-BN) and graphene lateral structure, a heterogeneous 2D structure, has been fabricated on single-crystal metals or metal foils by chemical vapor deposition (CVD). However, once fabricated on metals, the h-BN/graphene lateral structures require an additional transfer process for device applications, as reported for CVD graphene grown on metal foils. Here, we demonstrate that a single-crystal h-BN/graphene lateral structure can be epitaxially grown on a wide-gap semiconductor, SiC(0001). First, a single-crystal h-BN layer with the same orientation as bulk SiC was grown on a Si-terminated SiC substrate at 850 oC using borazine molecules. Second, when heated above 1150 oC in vacuum, the h-BN layer was partially removed and, subsequently, replaced with graphene domains. Interestingly, these graphene domains possess the same orientation as the h-BN layer, resulting in a single-crystal h-BN/graphene lateral structure on a whole sample area. For temperatures above 1600 oC, the single-crystal h-BN layer was completely replaced by the single-crystal graphene layer. The crystalline structure, electronic band structure, and atomic structure of the h-BN/graphene lateral structure were studied by using low energy electron diffraction, angle-resolved photoemission spectroscopy, and scanning tunneling microscopy, respectively. The h-BN/graphene lateral structure fabricated on a wide-gap semiconductor substrate can be directly applied to devices without a further transfer process, as reported for epitaxial graphene on a SiC substrate., Comment: 23 pages, 7 figures

We used two-dimensional Finte-Difference-Time-Domain (FDTD) software to study the transition behavior of nano-particles from scatterers to an optically uniform medium. We measured the transmission efficiency of the dipole source, which is located in the high refractive index medium(index=2.00) and encapsulated by low index resin(index=1.41). In an effort to compose index-matched resin and to reduce internal reflection, high-index nano-particles are added to low-index resin in simulations of various sizes and densities. As the size of the nano-particles and the average spacing between particles are reduced to 0.02 lambda and 0.07 lambda respectively, the transmission efficiency improves two-fold compared to that without nanoparticles. The numerical results can be used to understand the optical behavior of nano-particles and to improve the extraction efficiency of high brightness light-emitting-diodes(LEDs), through the use of nano-composite encapsulant., Comment: 9 pages, 5 jpg figures

A novel technique to measure heat transfer and liquid film thickness distributions over relatively large areas for two-phase flow and heat transfer phenomena using infrared (IR)thermometry is described. IR thermometry is an established technology that can be used to measure temperatures when optical access to the surface is available in the wavelengths of interest. In this work, a midwave IR camera (3.6-5.1 microns) is used to determine the temperature distribution within a multilayer consisting of a silicon substrate coated with a thin insulator. Since silicon is largely transparent to IR radiation, the temperature of the inner and outer walls of the multilayer can be measured by coating selected areas with a thin, IR opaque film. If the fluid used is also partially transparent to IR, the flow can be visualized and the liquid film thickness can be measured. The theoretical basis for the technique is given along with a description of the test apparatus and data reduction procedure. The technique is demonstrated by determining the heat transfer coefficient distributions produced by droplet evaporation and flow boiling heat transfer.