Search

We provide in this paper a comprehensive comparison of various transfer learning strategies and deep learning architectures for computer-aided classification of adult-type diffuse gliomas. We evaluate the generalizability of out-of-domain ImageNet representations for a target domain of histopathological images, and study the impact of in-domain adaptation using self-supervised and multi-task learning approaches for pretraining the models using the medium-to-large scale datasets of histopathological images. A semi-supervised learning approach is furthermore proposed, where the fine-tuned models are utilized to predict the labels of unannotated regions of the whole slide images (WSI). The models are subsequently retrained using the ground-truth labels and weak labels determined in the previous step, providing superior performance in comparison to standard in-domain transfer learning with balanced accuracy of 96.91% and F1-score 97.07%, and minimizing the pathologist's efforts for annotation. Finally, we provide a visualization tool working at WSI level which generates heatmaps that highlight tumor areas; thus, providing insights to pathologists concerning the most informative parts of the WSI.

The basal ganglia (BG) show a variety of functions for motor and cognition. There are two competitive pathways in the BG; direct pathway (DP) which facilitates movement and indirect pathway (IP) which suppresses movement. It is well known that diverse functions of the BG may be made through "balance" between DP and IP. But, to the best of our knowledge, so far no quantitative analysis for such balance was done. In this paper, as a first time, we introduce the competition degree ${\cal C}_d$ between DP and IP. Then, by employing ${\cal C}_d$, we quantify their competitive harmony (i.e., competition and cooperative interplay), which could lead to improving our understanding of the traditional "balance" so clearly and quantitatively. We first consider the case of normal dopamine (DA) level of $\phi^*=0.3$. In the case of phasic cortical input (10 Hz), a healthy state with ${\cal C}_d^* = 2.82$ (i.e., DP is 2.82 times stronger than IP) appears. In this case, normal movement occurs via harmony between DP and IP. Next, we consider the case of decreased DA level, $\phi = \phi^*(=0.3)~x_{DA}$ ($1 > x_{DA} \geq 0$). With decreasing $x_{DA}$ from 1, the competition degree ${\cal C}_d$ between DP and IP decreases monotonically from ${\cal C}_d^*$, which results in appearance of a pathological Parkinsonian state with reduced ${\cal C}_d$. In this Parkinsonian state, strength of IP is much increased than that in the case of normal healthy state, leading to disharmony between DP and IP. Due to such break-up of harmony between DP and IP, impaired movement occurs. Finally, we also study treatment of the pathological Parkinsonian state via recovery of harmony between DP and IP.

We experimentally explored novel behaviors of non-reciprocal absorption and almost zero reflection in a dual photon resonator system, which is physically separated and composed of two inverted split ring resonators (ISRRs) with varying inter-distances. We also found that an electromagnetically-induced-transparency (EIT)-like peak at a specific inter-distance of d = 18 mm through traveling waves flowing along a shared microstrip line to which the dual ISRRs are dissipatively coupled. With the aid of CST-simulations and analytical modeling, we found that destructive and/or constructive interferences in traveling waves, indirectly coupled to each ISRR, result in a traveling-wave-induced transparency peak within a narrow window. Furthermore, we observed not only strong non-reciprocal responses of reflectivity and absorptivity at individual inter-distances exactly at the corresponding EIT-like peak positions, but also nearly zero reflection and almost perfect absorption for a specific case of d = 20 mm. Finally, the unidirectional absorptions with zero reflection at d = 20 mm are found to be ascribed to a non-Hermitian origin. This work not only provides a better understanding of traveling-wave-induced indirect coupling between two photonic resonators without magnetic coupling, but also suggests potential implications for the resulting non-reciprocal behaviors of absorption and reflection in microwave circuits and quantum information devices.

Hypergraphs can naturally model group-wise relations (e.g., a group of users who co-purchase an item) as hyperedges. Hyperedge prediction is to predict future or unobserved hyperedges, which is a fundamental task in many real-world applications (e.g., group recommendation). Despite the recent breakthrough of hyperedge prediction methods, the following challenges have been rarely studied: (C1) How to aggregate the nodes in each hyperedge candidate for accurate hyperedge prediction? and (C2) How to mitigate the inherent data sparsity problem in hyperedge prediction? To tackle both challenges together, in this paper, we propose a novel hyperedge prediction framework (CASH) that employs (1) context-aware node aggregation to precisely capture complex relations among nodes in each hyperedge for (C1) and (2) self-supervised contrastive learning in the context of hyperedge prediction to enhance hypergraph representations for (C2). Furthermore, as for (C2), we propose a hyperedge-aware augmentation method to fully exploit the latent semantics behind the original hypergraph and consider both node-level and group-level contrasts (i.e., dual contrasts) for better node and hyperedge representations. Extensive experiments on six real-world hypergraphs reveal that CASH consistently outperforms all competing methods in terms of the accuracy in hyperedge prediction and each of the proposed strategies is effective in improving the model accuracy of CASH. For the detailed information of CASH, we provide the code and datasets at: https://github.com/yy-ko/cash., Comment: 12 pages, 11 figures

We explore magnetic-field-driven chaos in magnetic skyrmions. Oscillating magnetic fields induce nonlinear dynamics in skyrmions, arising from the coupling of the secondary gyrotropic mode with a non-uniform, breathing-like mode. Through micromagnetic simulations, we observe complex patterns of hypotrochoidal motion in the orbital trajectories of the skyrmions, which are interpreted using bifurcation diagrams and local Lyapunov exponents. Our findings demonstrate that different nonlinear behaviors of skyrmions emerge at distinct temporal stages, depending on the nonlinear dynamic parameters. Investigating the abundant dynamic patterns of skyrmions during the emergence of chaos not only enhances device reliability but also provides useful guidelines for establishing chaos computing based on skyrmion dynamics.

The problem of representing nodes in a signed network as low-dimensional vectors, known as signed network embedding (SNE), has garnered considerable attention in recent years. While several SNE methods based on graph convolutional networks (GCN) have been proposed for this problem, we point out that they significantly rely on the assumption that the decades-old balance theory always holds in the real-world. To address this limitation, we propose a novel GCN-based SNE approach, named as TrustSGCN, which corrects for incorrect embedding propagation in GCN by utilizing the trustworthiness on edge signs for high-order relationships inferred by the balance theory. The proposed approach consists of three modules: (M1) generation of each node's extended ego-network; (M2) measurement of trustworthiness on edge signs; and (M3) trustworthiness-aware propagation of embeddings. Furthermore, TrustSGCN learns the node embeddings by leveraging two well-known societal theories, i.e., balance and status. The experiments on four real-world signed network datasets demonstrate that TrustSGCN consistently outperforms five state-of-the-art GCN-based SNE methods. The code is available at https://github.com/kmj0792/TrustSGCN., Comment: 12 pages, 8 figures, 9 tables

Using both the in-out formalism and the monodromy method, we study the emission of charges from near-extremal charged Nariai black holes with the black hole and cosmological horizons close to each other, whose near-horizon geometry is $\mathrm{dS}_2 \times \mathrm{S}^2$. The emission becomes catastrophic for a charge with energy greater than its chemical potential, whose leading exponential factor increases inversely proportional to the separation of two horizons. This implies that near-extremal Nariai black holes with large charges quickly evaporate dominantly through the charge emission and evolve to black holes with a naked singularity, in contrast to near-extremal RN-dS black holes that have the Breitenlohner-Friedman bound below which they become stable against Hawking radiation and Schwinger effect of charge emission. The near-extremal Nariai black holes with small charges, which are close to near-extremal Schwarzschild-dS black holes, emit dominantly charge-neutral particles and evolve to black holes with increasing charge to mass ratio. We illuminate the origin of the catastrophic emission in the phase-integral formulation and monodromy method by comparing near-extremal charged Nariai black holes with near-extremal RN-dS black holes., Comment: revised version including corrections of mistakes and more physical discussions

Highly magnetized neutron stars have magnetic fields of order of the critical field and can lead to measurable QED effects. We consider the Goldreich-Julian pulsar model with supercritical magnetic fields, induced subcritical electric fields, and a period of milliseconds. We then study the strong field physics, such as Schwinger pair production and the vacuum birefringence including the wrench effect, whose X-ray polarimetry will be observed in future space missions., Comment: 13 pages, 6 figures; Talk at the Fifth Zeldovich meeting, an international conference in honor of Ya. B. Zeldovich held in Yerevan, Armenia on June 12--16, 2023

Sparse generalized matrix-matrix multiplication (SpGEMM) is a fundamental operation for real-world network analysis. With the increasing size of real-world networks, the single-machine-based SpGEMM approach cannot perform SpGEMM on large-scale networks, exceeding the size of main memory (i.e., not scalable). Although the distributed-system-based approach could handle large-scale SpGEMM based on multiple machines, it suffers from severe inter-machine communication overhead to aggregate results of multiple machines (i.e., not efficient). To address this dilemma, in this paper, we propose a novel storage-based SpGEMM approach (SAGE) that stores given networks in storage (e.g., SSD) and loads only the necessary parts of the networks into main memory when they are required for processing via a 3-layer architecture. Furthermore, we point out three challenges that could degrade the overall performance of SAGE and propose three effective strategies to address them: (1) block-based workload allocation for balancing workloads across threads, (2) in-memory partial aggregation for reducing the amount of unnecessarily generated storage-memory I/Os, and (3) distribution-aware memory allocation for preventing unexpected buffer overflows in main memory. Via extensive evaluation, we verify the superiority of SAGE over existing SpGEMM methods in terms of scalability and efficiency., Comment: 11 pages, 9 figures, 5 tables, CIKM

In communication and storage systems, error correction codes (ECCs) are pivotal in ensuring data reliability. As deep learning's applicability has broadened across diverse domains, there is a growing research focus on neural network-based decoders that outperform traditional decoding algorithms. Among these neural decoders, Error Correction Code Transformer (ECCT) has achieved the state-of-the-art performance, outperforming other methods by large margins. To further enhance the performance of ECCT, we propose two novel methods. First, leveraging the systematic encoding technique of ECCs, we introduce a new masking matrix for ECCT, aiming to improve the performance and reduce the computational complexity. Second, we propose a novel transformer architecture of ECCT called a double-masked ECCT. This architecture employs two different mask matrices in a parallel manner to learn more diverse features of the relationship between codeword bits in the masked self-attention blocks. Extensive simulation results show that the proposed double-masked ECCT outperforms the conventional ECCT, achieving the state-of-the-art decoding performance with significant margins., Comment: 8 pages, 5 figures

Introducing quasiparticle anisotropy in graphene via uniaxial strain has a profound effect on the polarization charge density induced by external impurities, both Coulomb and short-range. In particular, the charge distribution induced by a Coulomb impurity exhibits a power law tail modulated by a strain-dependent admixture of angular harmonics. The appearance of distributed charge is in sharp contrast to the response in pristine/isotropic graphene, where for subcritical impurities the polarization charge is fully localized at the impurity position. It is also interesting to note that our results are obtained strictly at zero chemical potential, and the behavior is distinct from the familiar Friedel oscillations observed at finite chemical potential. We find that over a wide range of strain, the $d$-wave symmetry is dominant. The presence of Dirac cone tilt, relevant to some 2D materials beyond graphene, can also substantially affect the induced charge distribution. Finally we consider impurities with short range potentials, and study the effect of strain on the charge response. Our results were obtained in the continuum via perturbation theory valid for weak (subcritical) potentials, and supported by numerical lattice simulations based on density functional theory., Comment: 13 pages, 13 figures. Added new Section VI with new figures, and updated old figures

The lower energy peak of the spectral energy distribution of blazars has commonly been ascribed to synchrotron radiation from relativistic particles in the jets. Despite the consensus regarding jet emission processes, the particle acceleration mechanism is still debated. Here, we present the first X-ray polarization observations of PG 1553+113, a high-synchrotron-peak blazar observed by the Imaging X-ray Polarimetry Explorer (IXPE). We detect an X-ray polarization degree of $(10\pm2)\%$ along an electric-vector position angle of $\psi_X=86^{\circ}\pm8^{\circ}$. At the same time, the radio and optical polarization degrees are lower by a factor of $\sim$3. During our IXPE pointing, we observed the first orphan optical polarization swing of the IXPE era, as the optical angle of PG 1553+113 underwent a smooth monotonic rotation by about 125$^\circ$, with a rate of $\sim$17 degrees per day. We do not find evidence of a similar rotation in either radio or X-rays, which suggests that the X-ray and optically emitting regions are separate or, at most, partially co-spatial. Our spectro-polarimetric results provide further evidence that the steady-state X-ray emission in blazars originates in a shock-accelerated and energy-stratified electron population., Comment: Accepted for pubblication in The Astrophysical Journal Letters. 3 Figures, 13 pages

We search for quasi-stellar objects (QSOs) in a wide area of the south ecliptic pole (SEP) field, which has been and will continue to be intensively explored through various space missions. For this purpose, we obtain deep broadband optical images of the SEP field covering an area of $\sim$$14.5\times14.5$ deg$^2$ with the Korea Microlensing Telescope Network. The 5$\sigma$ detection limits for point sources in the $BVRI$ bands are estimated to be $\sim$22.59, 22.60, 22.98, and 21.85 mag, respectively. Utilizing data from Wide-field Infrared Survey Explorer, unobscured QSO candidates are selected among the optically point-like sources using the mid-infrared (MIR) and optical-MIR colors. To further refine our selection and eliminate any contamination not adequately removed by the color-based selection, we perform the spectral energy distribution fitting with archival photometric data ranging from optical to MIR. As a result, we identify a total of 2,383 unobscured QSO candidates in the SEP field. We also apply a similar method to the north ecliptic pole field using the Pan-STARRS data and obtain a similar result of identifying 2,427 candidates. The differential number count per area of our QSO candidates is in good agreement with those measured from spectroscopically confirmed ones in other fields. Finally, we compare the results with the literature and discuss how this work will be implicated in future studies, especially with the upcoming space missions., Comment: 14 pages, 9 figures, accepted for publication in ApJS

DeepLab is a widely used deep neural network for semantic segmentation, whose success is attributed to its parallel architecture called atrous spatial pyramid pooling (ASPP). ASPP uses multiple atrous convolutions with different atrous rates to extract both local and global information. However, fixed values of atrous rates are used for the ASPP module, which restricts the size of its field of view. In principle, atrous rate should be a hyperparameter to change the field of view size according to the target task or dataset. However, the manipulation of atrous rate is not governed by any guidelines. This study proposes practical guidelines for obtaining an optimal atrous rate. First, an effective receptive field for semantic segmentation is introduced to analyze the inner behavior of segmentation networks. We observed that the use of ASPP module yielded a specific pattern in the effective receptive field, which was traced to reveal the module's underlying mechanism. Accordingly, we derive practical guidelines for obtaining the optimal atrous rate, which should be controlled based on the size of input image. Compared to other values, using the optimal atrous rate consistently improved the segmentation results across multiple datasets, including the STARE, CHASE_DB1, HRF, Cityscapes, and iSAID datasets., Comment: 18 pages, 12 figures

Accurate interpolation of seismic data is crucial for improving the quality of imaging and interpretation. In recent years, deep learning models such as U-Net and generative adversarial networks have been widely applied to seismic data interpolation. However, they often underperform when the training and test missing patterns do not match. To alleviate this issue, here we propose a novel framework that is built upon the multi-modal adaptable diffusion models. In the training phase, following the common wisdom, we use the denoising diffusion probabilistic model with a cosine noise schedule. This cosine global noise configuration improves the use of seismic data by reducing the involvement of excessive noise stages. In the inference phase, we introduce the denoising diffusion implicit model to reduce the number of sampling steps. Different from the conventional unconditional generation, we incorporate the known trace information into each reverse sampling step for achieving conditional interpolation. To enhance the coherence and continuity between the revealed traces and the missing traces, we further propose two strategies, including successive coherence correction and resampling. Coherence correction penalizes the mismatches in the revealed traces, while resampling conducts cyclic interpolation between adjacent reverse steps. Extensive experiments on synthetic and field seismic data validate our model's superiority and demonstrate its generalization capability to various missing patterns and different noise levels with just one training session. In addition, uncertainty quantification and ablation studies are also investigated., Comment: 14 pages, 13 figures