Massive Background Research Articles

Focus-Attention Approach in Optimizing DETR for Object Detection from High-Resolution Images

Transformer-based detectors have recently achieved remarkable success in object detection, revolutionizing the field with their efficiency and accuracy. However, applying these models to high-resolution images presents significant challenges due to the increased computational demands and complexity of processing dense, high-resolution data. In this paper, we introduce a novel model specifically designed for object detection from high-resolution imagery. This model incorporates a multi-layer object-focus network along with a transformer encoder-decoder structure. Specifically, the model employs a dual-head strategy in the object-focus network to balance the detailed analysis of small objects with computational efficiency. This is achieved by leveraging data sparsity to reduce unnecessary computations in massive background regions. Additionally, to improve detection performance for small objects, we propose a method to effectively apply the transformer encoder-decoder structure combined with the object-focus network on the multi-layer feature maps of the feature pyramid. Our extensive evaluations of the VisDrone, MS-COCO, and UAVid datasets demonstrate that our model outperforms other DETR-based detectors in both detection accuracy and computational speed, highlighting its superior performance. These results indicate a significant advancement in the field of high-resolution object detection utilizing transformer architecture.

Effects of massive gravity on s-wave holographic superconductor

In this paper, analytical investigation of the properties of [Formula: see text]-wave holographic superconductors in the background of a massive gravity theory in the probe limit has been carried out by employing the Sturm–Liouville eigenvalue method. We obtain the analytical expression for the relation between the critical temperature and the charge density. We also obtain the expression for the condensation operator and value of the critical exponent. We observe that as we increase the massive gravity couplings, the critical temperature increases and the condensate decreases. Then we compute the frequency dependence of conductivity by solving analytically the wave equation for electromagnetic perturbations. From the real part of the conductivity, we finally estimate the energy band gap. Our results show that as one keeps on increasing the coupling parameters of the massive gravity background, the band gap energy increases compared to the holographic superconductors constructed in the Einstein gravity background. The results indicate that massive background is more favorable than Einstein gravity background for constructing a gravity dual of the strongly coupled high [Formula: see text] superconductor as it enhances the value of the critical temperature.

Class-aware tiny object recognition over large-scale 3D point clouds

Although tremendous strides have been made in object recognition over large-scale 3D point clouds, one of the remaining open challenges is detecting tiny objects. The signal and appearance information of tiny objects are generally sparse and insufficient due to their tiny size, which makes the tiny object difficult to be detected with existing studies both in tiny 2D object recognition and general 3D object recognition since 3D topology relations are lost due to feature projection in tiny 2D object recognition and massive background may dominate the feature learner in general 3D object recognition. To the best of our knowledge, we explore tiny object recognition for the first time over large-scale 3D point clouds by designing deep neural network. To cope with this problem, a novel two-stage approach is proposed in this paper, which can effectively learn an informative and discriminative feature representation and efficiently process large-scale point clouds. Specifically, a class-aware filtering strategy is designed to dispense with the redundant background information. After that, we introduce a novel aggregating and sampling scheme in a supervised manner to progressively increase the receptive field for each 3D point, which can adaptively capture useful signals from tiny objects and learn complex geometric structures. Last, we propose a biased loss function that is inversely correlated to the number of tiny object points, thereby effectively handling the sparsity of tiny objects. Extensive experimental results over two real-world data sets validate the effectiveness and efficiency of the proposed approach.

High-Throughput Separation of Long DNA in Deterministic Lateral Displacement Arrays.

Length-based separation of DNA remains as relevant today as when gel electrophoresis was introduced almost 100 years ago. While new, long-read genomics technologies have revolutionised accessibility to powerful genomic data, the preparation of samples has not proceeded at the same pace, with sample preparation often constituting a considerable bottleneck, both in time and difficulty. Microfluidics holds great potential for automated, sample-to-answer analysis via the integration of preparatory and analytical steps, but for this to be fully realised, more versatile, powerful and integrable unit operations, such as separation, are essential. We demonstrate the displacement and separation of DNA with a throughput that is one to five orders of magnitude greater than other microfluidic techniques. Using a device with a small footprint (23 mm × 0.5 mm), and with feature sizes in the micrometre range, it is considerably easier to fabricate than parallelized nano-array-based approaches. We show the separation of 48.5 kbp and 166 kbp DNA strands achieving a significantly improved throughput of 760 ng/h, compared to previous work and the separation of low concentrations of 48.5 kbp DNA molecules from a massive background of sub 10 kbp fragments. We show that the extension of DNA molecules at high flow velocities, generally believed to make the length-based separation of long DNA difficult, does not place the ultimate limitation on our method. Instead, we explore the effects of polymer rotations and intermolecular interactions at extremely high DNA concentrations and postulate that these may have both negative and positive influences on the separation depending on the detailed experimental conditions.

Polyakov loop, gluon mass, gluon condensate, and its asymmetry near deconfinement

We consider a BRST-invariant generalization of the ``massive background Landau gauge,'' resembling the original Curci-Ferrari model that saw a revived interest due to its phenomenological success in modeling infrared Yang-Mills dynamics, including that of the phase transition. Unlike the Curci-Ferrari model, however, the mass parameter is no longer a phenomenological input, but it enters as a result of dimensional transmutation via a BRST-invariant dimension-2 gluon condensate. The associated renormalization constant is dealt with using Zimmermann's reduction of constants program, which fixes the value of the mass parameter to values close to those obtained within the Curci-Ferrari approach. Using a self-consistent background field, we can include the Polyakov loop and probe the deconfinement transition, including its interplay with the condensate and its electric--magnetic asymmetry. We report a continuous phase transition at ${T}_{c}\ensuremath{\approx}0.230\text{ }\text{ }\mathrm{GeV}$ in the SU(2) case and a first-order one at ${T}_{c}\ensuremath{\approx}0.164\text{ }\text{ }\mathrm{GeV}$ in the SU(3) case, values which are again rather close to those obtained within the Curci-Ferrari model at one-loop order.

Quantum Fisher information as a probe for Unruh thermality

A long-standing debate on Unruh effect is about its obscure thermal nature. In this Letter, we use quantum Fisher information (QFI) as an effective probe to explore the thermal nature of Unruh effect from both local and global perspectives. By resolving the full dynamics of UDW detector, we find that the QFI is a time-evolving function of detector's energy gap, Unruh temperature TU and particularities of background field, e.g., mass and spacetime dimensionality. We show that the asymptotic QFI whence detector arrives its equilibrium is solely determined by TU, demonstrating the global side of Unruh thermality alluded by the KMS condition. We also show that the local side of Unruh effect, i.e., the different ways for the detector to approach the same thermal equilibrium, is encoded in the corresponding time-evolution of the QFI. In particular, we find that with massless scalar background the QFI has unique monotonicity in n=3 dimensional spacetime, and becomes non-monotonous for n≠3 models where a local peak value exists at early time and for finite acceleration, indicating an enhanced precision of estimation on Unruh temperature at a relative low acceleration can be achieved. Once the field acquiring mass, the related QFI becomes significantly robust against the Unruh decoherence in the sense that its local peak sustains for a very long time. While coupling to a more massive background, the persistence can even be strengthened and the QFI possesses a larger maximal value. Such robustness of QFI can surely facilitate any practical quantum estimation task.

A comparison on predicting functional impact of genomic variants.

ABSTRACTSingle-nucleotide polymorphism (SNPs) may cause the diverse functional impact on RNA or protein changing genotype and phenotype, which may lead to common or complex diseases like cancers. Accurate prediction of the functional impact of SNPs is crucial to discover the ‘influential’ (deleterious, pathogenic, disease-causing, and predisposing) variants from massive background polymorphisms in the human genome. Increasing computational methods have been developed to predict the functional impact of variants. However, predictive performances of these computational methods on massive genomic variants are still unclear. In this regard, we systematically evaluated 14 important computational methods including specific methods for one type of variant and general methods for multiple types of variants from several aspects; none of these methods achieved excellent (AUC ≥ 0.9) performance in both data sets. CADD and REVEL achieved excellent performance on multiple types of variants and missense variants, respectively. This comparison aims to assist researchers and clinicians to select appropriate methods or develop better predictive methods.

Cardiac MRI segmentation with focal loss constrained deep residual networks

Delineating anatomical structures for cardiac magnetic resonance imaging (CMRI) is crucial for various medical applications such as medical diagnoses, treatment, and pathological studies. CMRI segmentation, which aims to automatically and accurately segment the heart structures, is highly beneficial for cardiologists. However, it is non-trivial to perfectly segment the ventricles, especially for the heart apex slices, considering their small sizes compared to the input images. For example, the endocardium in the Sunnybrook dataset only occupies 4% of the entire image by average. During the training process, these target pixels, or other hard samples, are buried by the massive backgrounds that make the model mostly receive optimization signals from easy samples. In this paper, we propose a focal loss constrained residual network (FR-Net) to tackle the problem. In order to mitigate the fact that the gradients of the hard samples can be easily overwhelmed by the easy samples, we use a pixel-wise re-weighting strategy to balance the gradients. Furthermore, considering focal loss constraints for each pixel independently, we propose an alternative training fashion that trains the model with focal loss and dice loss alternatively. The segmentation model can not only benefit from the pixel-wise focal loss but also from the region-wise dice loss to comprehensively optimize the model. We conducted thorough experiments on the Sunnybrook dataset, CMRI dataset, right ventricle dataset, and ACDC dataset to verify the effectiveness of the proposed method.

Megahertz repetition rate satellite laser ranging demonstration at Graz observatory.

Using ultra-high repetition rate lasers (≥100kHz) is one of the most promising strategies for the next generation of satellite laser ranging (SLR) systems. We present successful 1 MHz repetition rate SLR to targets up to inclined geosynchronous orbits at nighttime. Among those, a maximum return rate of up to 53% was achieved, equivalent to 265 k returns per second for the satellite Swarm-B. In addition, daytime megahertz (MHz) SLR was realized by utilizing a propagated MHz range gate to reduce the massive background noise. In the future, MHz SLR will greatly improve current technology with respect to data amount and data precision, shorter acquisition time, target signature detection, and attitude determination.

Feature Rescaling and Fusion for Tiny Object Detection

Recent years have witnessed rapid developments on computer vision, however, there are still challenges in detecting tiny objects in a large-scale background. The tiny objects knowledge become sparse and weak due to their tiny size, which makes the tiny objects difficult to be detected with the common approaches. In this paper, a new network named Specific Characteristics based Feature Rescaling and Fusion (SFRF) is designed to detect tiny persons in a broad horizon and massive background. Different from the methods in general, a Nonparametric Adaptive Dense Perceiving Algorithm (NADPA) is designed to automatically select and generate a new resized feature map with the high density distribution of tiny objects. Then, a method called Many-For-One strategy is used for feature fusion of the feature pyramid network (FPN) layers to improve the feature representation and detection. Finally, an ensemble model method named hierarchical Coarse-to-fine mechanism is designed based on the proposed methods to further improve the performance. The experiments demonstrate that the proposed approach achieves an obvious performance improvement on tiny object detection than the existing approaches, and our approach has been awarded as the 1st-place in the first large-scale Tiny Object Detection (TOD) challenge.

Improved Flow Cytometric Light Scatter Detection of Submicron-Sized Particles by Reduction of Optical Background Signals.

Flow cytometry allows multiparameter analysis on a single‐cell basis and is currently the method of choice to rapidly assess heterogeneity of cell populations in suspension. With the research field of extracellular vesicles (EV) rapidly expanding, there is an increased demand to address heterogeneity of EV populations in biological samples. Although flow cytometry would be the ideal technique to do so, the available instruments are in general not equipped to optimally detect the dim light scatter signals generated by submicron‐sized particles like EV. Although sideward scatter light and fluorescence are currently used as a threshold signal to identify EV within samples, the forward scatter light (FSC) parameter is often neglected due to the lack of resolution to distinguish EV‐related signals from noise. However, after optimization of FSC detection by adjusting the size of the obscuration bar, we recently showed that certain EV‐subsets could only be identified based on FSC. This observation made us to further study the possibilities to enhance FSC‐detection of submicron‐sized particles. By testing differently sized obscuration bars and differently sized pinholes in the focal plane behind the FSC detection lens, we generated a matrix that allowed us to determine which combination resulted in the lowest optical background in terms of numbers of events regarding FSC detection of submicron‐sized particles. We found that a combination of an 8‐mm obscuration bar and a 200‐μm pinhole reduced optical background in a reproducible manner to such extent that it allowed a robust separation of 100‐nm polystyrene beads from background signals within the FSC channel, and even allowed thresholding on FSC without the interference of massive background signals when both beads and EV were measured. These technical adaptations thus significantly improved FSC detection of submicron‐sized particles and provide an important lead for the further development and design of flow cytometers that aid in detection of submicron‐sized particles. © 2020 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

Protein Structure Determination in Living Cells.

To date, in-cell NMR has elucidated various aspects of protein behaviour by associating structures in physiological conditions. Meanwhile, current studies of this method mostly have deduced protein states in cells exclusively based on ‘indirect’ structural information from peak patterns and chemical shift changes but not ‘direct’ data explicitly including interatomic distances and angles. To fully understand the functions and physical properties of proteins inside cells, it is indispensable to obtain explicit structural data or determine three-dimensional (3D) structures of proteins in cells. Whilst the short lifetime of cells in a sample tube, low sample concentrations, and massive background signals make it difficult to observe NMR signals from proteins inside cells, several methodological advances help to overcome the problems. Paramagnetic effects have an outstanding potential for in-cell structural analysis. The combination of a limited amount of experimental in-cell data with software for ab initio protein structure prediction opens an avenue to visualise 3D protein structures inside cells. Conventional nuclear Overhauser effect spectroscopy (NOESY)-based structure determination is advantageous to elucidate the conformations of side-chain atoms of proteins as well as global structures. In this article, we review current progress for the structure analysis of proteins in living systems and discuss the feasibility of its future works.

AdS2 \xd7 S7 solutions from D0-F1-D8 intersections

We study an exhaustive analytic class of massive type IIA backgrounds pre-serving sixteen real supercharges and enjoying SL(2, ℝ) × SO(8) bosonic symmetry. The corresponding geometry is described by AdS2 × S7 warped over a line, which turns out to emerge from taking the near-horizon limit of D0-F1-D8 intersections. By studying the singularity structure of these solutions we find the possible presence of localized O8/D8 sources, as well as of fundamental strings smeared over the S7. Finally we discuss the relation between the aforementioned solutions and the known AdS7 × S2 class through double analytic continuation.

The ATLAS Run-2 Trigger Menu for Higher Luminosities: Design, Performance and Operational Aspects

The ATLAS experiment aims at recording about 1 kHz of physics collisions, starting with an LHC design bunch crossing rate of 40 MHz. To reduce the massive background rate while maintaining a high selection effciency for rare physics events (such as beyond the Standard Model physics), a two-level trigger system is used. Events are selected based on physics signatures such as the presence of energetic leptons, photons, jets or large missing energy. The trigger system exploits geometrical information on candidate objects, as well as multi-variate methods to carry out the necessary physics filtering. In total, the ATLAS online selection consists of thousands of different individual triggers. A trigger menu is a compilation of these triggers which specifies the physics algorithms to be used during data taking and the bandwidth a given trigger is allocated to. Trigger menus reflect not only the physics goals of the collaboration for a given run, but also take into consideration the instantaneous luminosity of the LHC and limitations from the ATLAS detector readout and online processing farm. For the 2017 run, the ATLAS trigger has been enhanced to be able to handle higher instantaneous luminosities (up to 2:0x1034 cm-2s-1) and to ensure the selection robustness against higher average multiple interactions per bunch crossing. In these proceedings, we describe the design criteria for the trigger menus used for Run 2 at the LHC. We discuss several aspects of the process, from the validation of the algorithms, the fine-tuning of the prescales, and the monitoring tools that ensure the smooth operation of the trigger during data taking. We also report on the physics performance of a few trigger algorithms.

Penrose limits and spin chains in the GJV/CS-SYM duality

We examine Penrose limits of the duality proposed by Guarino, Jafferis and Varela between a type IIA massive background of the type of a warped, squashed AdS4 × S6, and a 2+1 dimensional IR fixed point of mathcal{N}=8 super Yang-Mills deformed by Chern-Simons terms to mathcal{N}=2 supersymmetry. One type of Penrose limit for closed strings corresponds to a large charge closed spin chain, and another, for open strings on giant graviton D-branes, corresponds to an open spin chain on sub-determinant operators. For the first limit, we find that like in the ABJM case, there are functions fa(λ) that interpolate between the perturbative and nonperturbative (string) regions for the magnon energy. For the second, we are unable to match the gravity result with the expected field theory result, making this model more interesting than ones with more supersymmetry.

DOMINO: Using Machine Learning to Predict Genes Associated with Dominant Disorders

In contrast to recessive conditions with biallelic inheritance, identification of dominant (monoallelic) mutations for Mendelian disorders is more difficult, because of the abundance of benign heterozygous variants that act as massive background noise (typically, in a400:1 excess ratio). To reduce this overflow of false positives in next-generation sequencing (NGS) screens, we developed DOMINO, a tool assessing the likelihood for a gene to harbor dominant changes. Unlike commonly-used predictors of pathogenicity, DOMINO takes into consideration features that are the properties of genes, rather than of variants. It uses a machine-learning approach to extract discriminant information from a broad array of features (N = 432), including: genomic data, intra-, and interspecies conservation, gene expression, protein-protein interactions, protein structure, etc. DOMINO's iterative architecture includes a training process on 985 genes with well-established inheritance patterns for Mendelian conditions, and repeated cross-validation that optimizes its discriminant power. When validated on 99 newly-discovered genes with pathogenic mutations, the algorithm displays an excellent final performance, with an area under the curve (AUC) of 0.92. Furthermore, unsupervised analysis by DOMINO of real sets of NGS data from individuals with intellectual disability or epilepsy correctly recognizes known genes and predicts 9 new candidates, with very high confidence. In summary, DOMINO is a robust and reliable tool that can infer dominance of candidate genes with high sensitivity and specificity, making it a useful complement to any NGS pipeline dealing with the analysis of the morbid human genome.

Detection of trace amounts of target DNA from massive background of nucleic acids by using the LM-PCR-based preamplification method.

The sensitivity and specificity of DNA detection may decrease when the target DNA is in very low abundance. To effectively detect trace amounts of target DNA from massive background of nucleic acids, we have developed a powerful multiplex preamplification method based on ligation-mediated PCR that can greatly enrich multiple target DNAs from massive backgrounds. By employing type IIS restriction endonuclease (REase) and specifically designed oligonucleotide adapters, target DNA can be preamplified with high efficiency and sensitivity. Combining with normal PCR, 10 copies of target DNA was effectively detected from over 108 times more excessive backgrounds with high specificity and 10 times more effectively than conventional PCR. In particular, the usage of universal primer in the preamplification PCR (pre-amp PCR) ensured that multiple targets could be equivalently amplified, which was confirmed by quantitative PCR (qPCR), indicating it could meet the demands of high-throughput detection. The flexibility and applicability of pre-amp PCR was validated by using different microorganisms DNA as targets and employing two different type IIS REases. The results suggest that the pre-amp PCR method has broad application prospects in various gene detection fields.

Simultaneous Identification of Multiple Causal Mutations in Rice.

Next-generation sequencing technologies (NGST) are being used to discover causal mutations in ethyl methanesulfonate (EMS)-mutagenized plant populations. However, the published protocols often deliver too many candidate sites and sometimes fail to find the mutant gene of interest. Accurate identification of the causal mutation from massive background polymorphisms and sequencing deficiencies remains challenging. Here we describe a NGST-based method, named SIMM, that can simultaneously identify the causal mutations in multiple independent mutants. Multiple rice mutants derived from the same parental line were back-crossed, and for each mutant, the derived F2 individuals of the recessive mutant phenotype were pooled and sequenced. The resulting sequences were aligned to the Nipponbare reference genome, and single nucleotide polymorphisms (SNPs) were subsequently compared among the mutants. Allele index (AI) and Euclidean distance (ED) were incorporated into the analysis to reduce noises caused by background polymorphisms and re-sequencing errors. Corrections of sequence bias against GC- and AT-rich sequences in the candidate region were conducted when necessary. Using this method, we successfully identified seven new mutant alleles from Huanghuazhan (HHZ), an elite indica rice cultivar in China. All mutant alleles were validated by phenotype association assay. A pipeline based on Perl scripts for SIMM is publicly available at https://sourceforge.net/projects/simm/.

Highly Sensitive Detection of Low-Abundance White Spot Syndrome Virus by a Pre-Amplification PCR Method.

White spot syndrome virus (WSSV) is a major threat to the shrimp farming industry and so far there is no effective therapy for it, and thus early diagnostic of WSSV is of great importance. However, at the early stage of infection, the extremely low-abundance of WSSV DNA challenges the detection sensitivity and accuracy of PCR. To effectively detect low-abundance WSSV, here we developed a pre-amplification PCR (pre-amp PCR) method to amplify trace amounts of WSSV DNA from massive background genomic DNA. Combining with normal specific PCR, 10 copies of target WSSV genes were detected from ~1010 magnitude of backgrounds. In particular, multiple target genes were able to be balanced amplified with similar efficiency due to the usage of the universal primer. The efficiency of the pre-amp PCR was validated by nested-PCR and quantitative PCR, and pre-amp PCR showed higher efficiency than nested-PCR when multiple targets were detected. The developed method is particularly suitable for the super early diagnosis of WSSV, and has potential to be applied in other low-abundance sample detection cases.

Highly specific DNA detection from massive background nucleic acids based on rolling circle amplification of target dsDNA

An advanced rolling circle amplification (RCA) strategy based on the target-circularization of the targeted double-stranded DNA (dsDNA) was established. Different from the traditional padlock-RCA, in which single-stranded probe DNA was circularized and amplified as signal amplification, our new approach could amplify a double-stranded DNA target. This special circularization of the target was realized by the ligation of target DNA with a biotin-labelled duplex adaptor containing 9 nt sticky ends by complementary base pairing. High specificity was obtained using two primers targeting the target sequence but not the probe itself in traditional padlock-RCA. With the help of streptavidin magnetic beads that immobilized the ligated dsDNA amplicon, the background nucleic acids contributing the most to non-specific amplification were eliminated. Under optimized conditions, less than 60 copies of the target sequence could be detected in the presence of massive background nucleic acids (>1012 copies of unrelated sequences). The sensitivity and specificity can rival canonical PCR. Without thermal cycles, the reduced handling and simpler equipment requirements render this assay a simple and rapid alternative to conventional methods. Based on these advantages, this method is a promising candidate in practical applications such as detecting contaminated food-borne pathogens in comprehensive food samples.