Search

With the rapid development of multimedia technology, considerable achievement has been achieved in image retrieval technology. On the one hand, users' demand for cross-modal retrieval is increasing rapidly. On the other hand, deep hashing algorithms as one of the most prominent high-dimensional reduction methods have received extensive attention. Under such background, the cross-modal retrieval method based on deep hashing came into being. Although the cross-modal hashing learning method has moved a long way in recent years, it still has space of promotion. First, existing cross-modal based hashing algorithms can only map the feature vectors into binary codes with a fixed length. The length of hash codes cannot be modified according to the practical situation. Second, the saturation level of the probability function used in existing methods is too high to concentrate relevant samples very well, which results in low retrieval efficiency. To solve the above problem, a novel variable-length hash code based on adaptive weight was proposed in this paper. The length of hash codes could be adjusted according to the importance of each bit. And a novel probability function based on Cauchy distribution was proposed to generate compact binary codes and make hashing retrieval more efficient. The experiment shows that the proposed cross-modal image retrieval algorithm based on deep hashing outperforms existing related algorithms on the accuracy of retrieval. [ABSTRACT FROM AUTHOR]

Motivated by the sequence reconstruction problem initiated by Levenshtein, reconstruction codes were introduced by Cai et al. to combat errors when a fixed number of noisy channels are available. The central problem on this topic is to design codes with sizes as large as possible, such that every codeword can be uniquely reconstructed from any N distinct noisy reads, where N is fixed. In this paper, we study binary reconstruction codes with the constraint that every codeword is balanced, which is a fundamental requirement in the technique of DNA-based storage. For all possible channels with a single edit error and their variants, we design asymptotically optimal balanced reconstruction codes for all N, and show that the number of their redundant symbols decreases from 3 2 log 2 n + O (1) to 1 2 log 2 n + log 2 log 2 n + O (1) , and finally to 1 2 log 2 n + O (1) but with different speeds, where n is the length of the code. Compared with the unbalanced case, our results imply that the balanced property does not reduce the rate of the reconstruction code in the corresponding codebook. [ABSTRACT FROM AUTHOR]

In this paper, we first consider the existence conditions, the construction and the enumeration of Hermitian self-dual θ -cyclic and θ -negacyclic codes over I F p 2 , where p is a prime number and θ is the Frobenius automorphism over I F p 2 . We then give necessary and sufficient conditions for the existence of Hermitian self-dual θ -cyclic and θ -negacyclic codes over I F p e where e is an even integer greater than 2. [ABSTRACT FROM AUTHOR]

Linear codes play a crucial role in various fields of engineering and mathematics, including data storage, communication, cryptography, and combinatorics. Minimal linear codes, a subset of linear codes, are particularly essential for designing effective secret sharing schemes. In this paper, we introduce several classes of minimal binary linear codes by carefully selecting appropriate Boolean functions. These functions belong to a renowned class of Boolean functions, namely, the general Maiorana–McFarland class. We employ a method first proposed by Ding et al. (IEEE Trans Inf Theory 64(10):6536–6545, 2018) to construct minimal codes violating the Ashikhmin–Barg bound (wide minimal codes) by using Krawtchouk polynomials. The lengths, dimensions, and weight distributions of the obtained codes are determined using the Walsh spectrum distribution of the chosen Boolean functions. Our findings demonstrate that a vast majority of the newly constructed codes are wide minimal. Furthermore, our proposed codes exhibit a significantly larger minimum distance, in some cases, compared to some existing similar constructions. Finally, we address this method, based on Krawtchouk polynomials, more generally, and highlight certain generic properties related to it. These general results offer insights into the scope of this approach. [ABSTRACT FROM AUTHOR]

In this paper, δ -dual codes over finite commutative semi-simple rings are defined as a generalization of dual codes over finite commutative semi-simple rings. Some properties of δ -dual codes are given. We present necessary and sufficient conditions for a λ -constacyclic code of length n to be δ -self-dual, δ -self-orthogonal, δ -dual-containing, δ -LCD over finite commutative semi-simple rings. We also study the δ -dual of skew Θ - λ -constacyclic codes over finite commutative semi-simple rings. Among others, we also give necessary and sufficient conditions for a skew Θ - λ -constacyclic code of any length n to be δ -self-dual, δ -self-orthogonal, δ -dual-containing, δ -LCD over finite commutative semi-simple rings. [ABSTRACT FROM AUTHOR]

The exponential growth of data raises an unprecedented challenge in data analysis: how to retrieve interesting information from such large-scale data. Hash learning is a promising solution to address this challenge, because it may bring many potential advantages, such as extremely high efficiency and low storage cost, after projecting high-dimensional data to compact binary codes. However, traditional hash learning algorithms often suffer from the problem of semantic inconsistency, where images with similar semantic features may have different binary codes. In this paper, we propose a novel end-to-end deep hashing method based on the similarities of binary codes, dubbed CSDH (Code Similarity-based Deep Hashing), for image retrieval. Specifically, it extracts deep features from images to capture semantic information using a pre-trained deep convolutional neural network. Additionally, a hidden and fully connected layer is attached at the end of the deep network to derive hash bits by virtue of an activation function. To preserve the semantic consistency of images, a loss function has been introduced. It takes the label similarities, as well as the Hamming embedding distances, into consideration. By doing so, CSDH can learn more compact and powerful hash codes, which not only can preserve semantic similarity but also have small Hamming distances between similar images. To verify the effectiveness of CSDH, we evaluate CSDH on two public benchmark image collections, i.e., CIFAR-10 and NUS-WIDE, with five classic shallow hashing models and six popular deep hashing ones. The experimental results show that CSDH can achieve competitive performance to the popular deep hashing algorithms. [ABSTRACT FROM AUTHOR]

Global contexts are critical to locating salient objects for salient object detection (SOD). However, the convolution operation in CNNs has a local receptive field, which cannot capture long-distance global information. Recent studies have shown that modernized CNN models with large kernel convolution, such as ConvNeXt, can effectively extend the receptive fields. Based on it, this paper explores the potential of large kernel CNN for SOD task. Inspired by the common information between RGB and depth images in salient objects, we propose a ConvNeXt-based Siamese network with shared weight parameters. This structural design can effectively reduce the number of parameters without sacrificing performance. Furthermore, a depth information preprocessing module is proposed to minimize the impact of low-quality depth images on predicted saliency maps. For cross-modal feature interaction, a dynamic fusion module is designed to enhance cross-modal complementarity dynamically. Extensive experiments and evaluation results on six benchmark datasets demonstrate the outstanding performance of the proposed method against 14 state-of-the-art RGB-D methods. Our code will be released at https://github.com/zyh5119232/CSNet. [ABSTRACT FROM AUTHOR]

In this paper, we give a classification of extremal Type II Z 4 -codes of length 24. This is equivalent to a classification of 4-frames of the Leech lattice up to the action of the automorphism group. [ABSTRACT FROM AUTHOR]

In this paper we construct distance-regular graphs admitting a vertex transitive action of the five sporadic simple groups discovered by E. Mathieu, the Mathieu groups M 11 , M 12 , M 22 , M 23 and M 24 . From the binary code spanned by an adjacency matrix of the strongly regular graph with parameters (176,70,18,34) we obtain block designs having the full automorphism groups isomorphic to the Higman-Sims finite simple group. Moreover, from that code we obtain eight 2-designs having the full automorphism group isomorphic to M 22 , whose existence cannot be explained neither by the Assmus-Mattson theorem nor by a transitivity argument. Further, we discuss a possibility of permutation decoding of the codes spanned by adjacency matrices of the graphs constructed and find small PD-sets for some of the codes. [ABSTRACT FROM AUTHOR]

Binary duadic codes are an interesting subclass of cyclic codes since they have large dimensions and their minimum distances may have a square-root bound. In this paper, we present several families of binary duadic codes of length 2 m - 1 and develop some lower bounds on their minimum distances by using the BCH bound on cyclic codes, which partially solves one case of the open problem proposed in Liu et al (Finite Field Appl 91:102270, 2023). It is shown that the lower bounds on their minimum distances are close to the square root bound. Moreover, the parameters of the dual and extended codes of these binary duadic codes are investigated. [ABSTRACT FROM AUTHOR]

CSS-T codes were recently introduced as quantum error-correcting codes that respect a transversal gate. A CSS-T code depends on a CSS-T pair, which is a pair of binary codes (C 1 , C 2) such that C 1 contains C 2 , C 2 is even, and the shortening of the dual of C 1 with respect to the support of each codeword of C 2 is self-dual. In this paper, we give new conditions to guarantee that a pair of binary codes (C 1 , C 2) is a CSS-T pair. We define the poset of CSS-T pairs and determine the minimal and maximal elements of the poset. We provide a propagation rule for nondegenerate CSS-T codes. We apply some main results to Reed–Muller, cyclic and extended cyclic codes. We characterize CSS-T pairs of cyclic codes in terms of the defining cyclotomic cosets. We find cyclic and extended cyclic codes to obtain quantum codes with better parameters than those in the literature. [ABSTRACT FROM AUTHOR]

In quantum key distribution (QKD), two users extract a shared secret key using a quantum communication channel in the presence of an eavesdropper. Among QKD protocols, the ones based on energy-time (ET) entanglement of photons have been studied extensively due to their ability to generate high key rates from the arrival times of entangled photons. For the information reconciliation (IR) stage of ET-QKD protocols (where the users communicate using a classical channel in order to reconcile differences in their data), a scheme called multi-level coding (MLC) was proposed by Zhou et al. in prior work. The MLC scheme splits the raw key symbols into bit layers and utilizes binary low-density parity check (LDPC) codes to encode each layer. Although binary LDPC codes are able to offer low complexity decoding for IR, they have poor error-correcting performance compared to their non-binary counterparts, thus leading to low key rates. Additionally, existing LDPC codes do not fully utilize the properties of the QKD channel to optimize the key rates. In this paper, we mitigate the above issues by proposing a flexible protocol for IR in ET-QKD systems called non-binary multi-level coding NB-MLC(a) which is parameterized by a positive integer a. The NB-MLC(a) protocol is a generalization of the MLC scheme and utilizes NB-LDPC codes from a Galois field of size 2 a . We show that by using a small value of a, the NB-MLC(a) protocol significantly improves the key rate without much increase in complexity. To further improve the key rates of the NB-MLC(a) protocol, we propose (i) a joint rate and degree distribution optimization (JRDO) algorithm to design the NB-LDPC codes for the protocol and (ii) an interleaved decoding and communication (IDC) scheme to decode the different layers of the NB-MLC(a) protocol. The JRDO algorithm is designed to use the QKD channel information, and we show that it results in a higher key rate than codes used in prior work. Additionally, the IDC scheme improves the key rate compared to the decoding and communication methods utilized previously in literature. Overall, the NB-MLC(a) protocol that uses JRDO-LDPC codes and the IDC scheme results in a significant 40–60 % improvement in key rates compared to prior work for ET-QKD systems. [ABSTRACT FROM AUTHOR]

Digital data transmission and storage are indispensable components for our contemporary information-driven society. As data volumes keep increasing, the need for efficient coding schemes grows. This paper gives a novel coding scheme for generating sixteen codes from generators with quaternary codes, thereby offering a significant increase in data representation capacity, new possibilities for data encoding and decoding, and an overall improvement in the efficiency of the encoding scheme. This provides up a new perspective for numerous fields that can profit from the proposed scheme, including coding theory, computer science, modeling, and experimental design. We focus on only experimental design and modeling, due to space limits and their importance in designing and modeling real-world and scientific phenomena. The main findings indicate that the new generated sixteen codes can be used as optimal experimental designs with good space-filling behavior when designing large-scale experiments, and as efficient datasets that perform well in fitting various functions even with a small number of points and without modeling assumptions. The relationships between the codewords of the generated sixteen codes and the codewords of the generator quaternary codes are investigated. These relationships provide a benchmark by which to select the generator most efficiently, resulting in codewords that are efficient from multiple angles. Furthermore, the space-filling behavior and isomorphism examination of the sixteen codes generated as designs are deeply theoretically studied. [ABSTRACT FROM AUTHOR]

In recent days, local binary pattern and their variants plays a vital role in classification of EEG signals. Hence, in this paper a novel method for classification of EEG signals in accordance with local binary pattern is proposed. Initially, the EEG signal with 9 points is considered and then the average of the signal points existing at various distances such as 1, 2 and 3 are computed. Then, the computed average value for an EEG segment is compared with the center signal point of an EEG segment and thus yields the binary value 0 or 1. Later, the majority rule is employed resulting in 8 bit binary pattern. Also, the average of EEG segment is compared with center point. The decimal equivalent of the generated binary code is referred as the proposed label. Later, the histogram is generated involving the proposed label and then this histogram are used as the feature for an EEG signal. The experimental results on Bonn and Freiburg dataset shows that the proposed method achieves 75.81% and 94.43%, 72.43% and 89.30% in sensitivity and specificity, respectively. Also, the performance is evaluated in terms of classification accuracy with KNN and SVMclassifier by changing the training and testing data. The results indicate that the proposed method could distinguish the seizure and seizure free EEG signals evidently. [ABSTRACT FROM AUTHOR]

This paper aims to explore the potential benefits and performance enhancements of combining a novel error correction code with polar codes through serial and parallel concatenation. The BGCC-EDAC (Binary to Gray Code Conversion Based Error Detection and Correction) code is a novel forward error correction code that functions as a channel coding technique. It has shown promising results in effectively detecting and correcting errors. This study presents an empirical analysis of the application of serial concatenation of BGCC-EDAC-aided polar code (SC-BPC) and parallel concatenation of BGCC-EDAC-aided polar code (PC-BPC). The simulation results obtained using Cadence 90 nm technology demonstrate that the SC-BPC and PC-BPC architectures presented in this study require fewer hardware resources compared to the other error correction codes currently available. The empirical analysis conducted using MATLAB demonstrates that the proposed codes exhibit a significant performance improvement compared to CRC-aided polar codes. The PC-BPC code proposed in this study exhibits a slight enhancement of 0.18 dB compared to the parallel concatenation technique of CRC-aided polar codes at a bit error rate of 10−4. Similarly, the SC-BPC code introduced in this research displays an identical improvement of 0.3dB when compared to the serial concatenation approach with CRC-aided polar codes at the same BER level. The PC-BPC and SC-BPC architectures are expected to be widely adopted in 5G wireless communication systems due to their advantageous features such as low latency, high throughput, and improved BER performance. [ABSTRACT FROM AUTHOR]

Background: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer.Objective: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer.Materials and methods: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer.Results: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer.Conclusion: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer.Graphical Abstract: Bone age assessment assists physicians in evaluating the growth and development of children. However, deep learning methods for bone age estimation do not currently incorporate differential features obtained through comparisons with other bone atlases.To propose a more accurate method, Delta-Age-Sex-AdaIn (DASA-net), for bone age assessment, this paper combines age and sex distribution through adaptive instance normalization (AdaIN) and style transfer, simulating the process of visually comparing hand images with a standard bone atlas to determine bone age.The proposed Delta-Age-Sex-AdaIn (DASA-net) consists of four modules: BoneEncoder, Binary code distribution, Delta-Age-Sex-AdaIn, and AgeDecoder. It is compared with state-of-the-art methods on both a public Radiological Society of North America (RSNA) pediatric bone age prediction dataset (14,236 hand radiographs, ranging from 1 to 228 months) and a private bone age prediction dataset from Zigong Fourth People's Hospital (474 hand radiographs, ranging from 12 to 218 months, 268 male). Ablation experiments were designed to demonstrate the necessity of incorporating age distribution and sex distribution.The DASA-net model achieved a lower mean absolute deviation (MAD) of 3.52 months on the RSNA dataset, outperforming other methods such as BoneXpert, Deeplasia, BoNet, and other deep learning based methods. On the private dataset, the DASA-net model obtained a MAD of 3.82 months, which is also superior to other methods.The proposed DASA-net model aided the model’s learning of the distinctive characteristics of hand bones of various ages and both sexes by integrating age and sex distribution into style transfer. [ABSTRACT FROM AUTHOR]

We study a class of integer bilevel programs with second-order cone constraints at the upper-level and a convex-quadratic objective function and linear constraints at the lower-level. We develop disjunctive cuts (DCs) to separate bilevel-infeasible solutions using a second-order-cone-based cut-generating procedure. We propose DC separation strategies and consider several approaches for removing redundant disjunctions and normalization. Using these DCs, we propose a branch-and-cut algorithm for the problem class we study, and a cutting-plane method for the problem variant with only binary variables. We present an extensive computational study on a diverse set of instances, including instances with binary and with integer variables, and instances with a single and with multiple linking constraints. Our computational study demonstrates that the proposed enhancements of our solution approaches are effective for improving the performance. Moreover, both of our approaches outperform a state-of-the-art generic solver for mixed-integer bilevel linear programs that is able to solve a linearized version of our binary instances. [ABSTRACT FROM AUTHOR]

In this work we define and study binary codes C q , k and C q , k ¯ obtained from neighborhood designs of Paley-type bipartite graphs P(q, k) and their complements, respectively for q an odd prime. We prove that for some values of q and k the codes C q , k are self-dual and the codes C q , k ¯ are self-orthogonal. Most of these codes tend to be with optimal or near optimal parameters. Next, we extend the codes C q , k to get doubly even self dual codes and find that most of these codes are extremal. [ABSTRACT FROM AUTHOR]

We give a new technique for constructing self-dual codes based on a block matrix whose blocks arise from group rings and orthogonal matrices. The technique can be used to construct self-dual codes over finite commutative Frobenius rings of characteristic 2. We give and prove the necessary conditions needed for the technique to produce self-dual codes. We also establish the connection between self-dual codes generated by the new technique and units in group rings. Using the construction together with the building-up construction, we obtain new extremal binary self-dual codes of lengths 64, 66 and 68 and new best known binary self-dual codes of length 80. [ABSTRACT FROM AUTHOR]

We present new constructions of binary quantum codes from quaternary linear Hermitian self-dual codes. Our main ingredients for these constructions are nearly self-orthogonal cyclic or duadic codes over F 4 . An infinite family of 0-dimensional binary quantum codes is provided. We give minimum distance lower bounds for our quantum codes in terms of the minimum distance of their ingredient linear codes. We also present new results on the minimum distance of linear cyclic codes using their fixed subcodes. Finally, we list many new record-breaking quantum codes obtained from our constructions. [ABSTRACT FROM AUTHOR]

In the field of image classification, few shot learning (FSL) is to identify the new samples for each category using the extremely limited data. Due to lacking of data, FSL was usually performed by optimizing network or searching for new measurement methods. However, these mechanisms may fail to obtain enough information from the limited test data and then lack of learning ability. To address this problem, we attempt to obtain more information from the unlabeled data using the pseudo labels. In order to improve the reliability of selecting pseudo label data, we propose a new sample selection strategy, named Metric Confidence Selection (MCS), which is more conducive to select the most reliable pseudo-label data. In addition, we propose a new framework to combine our MCS and metric learning together. Our framework tends to get more information from the unlabeled samples, which is helpful to improve utilization efficiency. Extensive experiments on four widely-used benchmark datasets show that our proposed method surpass most state-of-the-art ones in few shot image classification. [ABSTRACT FROM AUTHOR]

Anomaly detection for multivariate time series has been widely used in industry and has become one of the hot research problems in the field of data mining. However, the existing anomaly detection methods still have the following limitations: (1) The topological relationship between sensors is complex and nonlinear, and it is difficult to effectively model the inter-sensor dependencies. (2) Most anomaly detection models tend to ignore the critical temporal information in different time steps when capturing the temporal dependencies. To address these problems, we propose a hybrid graph transformer network for multivariate time series anomaly detection (HGTMAD), which combines the transformer with graph convolution to predict multivariate time series-based anomalies. In the time domain, we design a new sparse metric transformer network to capture the time dependence, where the Wasserstein distance is used to measure out significant dot product pairs to learn a better time series representation. In the spatial domain, we design a new channel convolution transformer network fused with a graph convolution network to learn accurately the complex dependencies of multivariate time series in spatial and on feature dimensions based on a combination of local and global approaches. Extensive experiments on publicly available datasets further demonstrate that the proposed HGTMAD significantly outperforms the mainstream state-of-the-art methods. [ABSTRACT FROM AUTHOR]

In this article, we construct linear codes over the commutative non-unital ring I of size four. We obtain their Lee-weight distributions and study their binary Gray images. Under certain mild conditions, these classes of binary codes are minimal and self-orthogonal. All codes in this article are few-weight codes. Besides, an infinite class of these binary codes consists of distance optimal codes with respect to the Griesmer bound. [ABSTRACT FROM AUTHOR]