Compact Code Research Articles

Method for scalable and performant GPU-accelerated simulation of multiphase compressible flow

Multiphase compressible flows are often characterized by a broad range of space and time scales, entailing large grids and small time steps. Simulations of these flows on CPU-based clusters can thus take several wall-clock days. Offloading the compute kernels to GPUs appears attractive but is memory-bound for many finite-volume and -difference methods, damping speedups. Even when realized, GPU-based kernels lead to more intrusive communication and I/O times owing to lower computation costs. We present a strategy for GPU acceleration of multiphase compressible flow solvers that addresses these challenges and obtains large speedups at scale. We use OpenACC for directive-based offloading of all compute kernels while maintaining low-level control when needed. An established Fortran preprocessor and metaprogramming tool, Fypp, enables otherwise hidden compile-time optimizations. This strategy exposes compile-time optimizations and high memory reuse while retaining readable, maintainable, and compact code. Remote direct memory access realized via CUDA-aware MPI and GPUDirect reduces halo-exchange communication time. We implement this approach in the open-source solver MFC [1]. Metaprogramming results in an 8-times speedup of the most expensive kernels compared to a statically compiled program, reaching 46% of peak FLOPs on modern NVIDIA GPUs and high arithmetic intensity (about 10 FLOPs/byte). In representative simulations, a single NVIDIA A100 GPU is 7-times faster compared to an Intel Xeon Cascade Lake (6248) CPU die, or about 300-times faster compared to a single such CPU core. At the same time, near-ideal (97%) weak scaling is observed for at least 13824 GPUs on OLCF Summit. A strong scaling efficiency of 84% is retained for an 8-times increase in GPU count. Collective I/O, implemented via MPI3, helps ensure the negligible contribution of data transfers (<1% of the wall time for a typical, large simulation). Large many-GPU simulations of compressible (solid-)liquid-gas flows demonstrate the practical utility of this strategy.

A Laplace operator-based active contour model with improved image edge detection performance

Active contour model (ACM) is an important branch in the field of image segmentation, since it adapts to different image types and scenes. However, traditional ACMs rely on local image information, which makes them sensitive to initial contours and time-consuming. In this paper, a Laplace operator based ACM (LOACM) with improved image edge detection performance is proposed. LOACM combines the global pre-fitting function which is constructed by using the image's second-order gradient information with the adaptive boundary indicator function to build a hybrid model. Its compact code size allows it to be executed on a wide range of edge devices. To substantiate the segmentation capability of the LOACM model, the BSD500 dataset was utilized for experimentation. The results indicate that LOACM exhibits excellent performance, with average IOU and DSC scores of 0.861 and 0.924 respectively. Additionally, it consistently handles processes within an average time of less than 0.3 s, significantly less than that of comparative models, providing compelling evidence of LOACM's exceptional efficiency and precision.

Rank-based Hashing for Effective and Efficient Nearest Neighbor Search for Image Retrieval

The large and growing amount of digital data creates a pressing need for approaches capable of indexing and retrieving multimedia content. A traditional and fundamental challenge consists of effectively and efficiently performing nearest-neighbor searches. After decades of research, several different methods are available, including trees, hashing, and graph-based approaches. Most of the current methods exploit learning to hash approaches based on deep learning. In spite of effective results and compact codes obtained, such methods often require a significant amount of labeled data for training. Unsupervised approaches also rely on expensive training procedures usually based on a huge amount of data. In this work, we propose an unsupervised data-independent approach for nearest neighbor searches, which can be used with different features, including deep features trained by transfer learning. The method uses a rank-based formulation and exploits a hashing approach for efficient ranked list computation at query time. A comprehensive experimental evaluation was conducted on 7 public datasets, considering deep features based on CNNs and Transformers. Both effectiveness and efficiency aspects were evaluated. The proposed approach achieves remarkable results in comparison to traditional and state-of-the-art methods. Hence, it is an attractive and innovative solution, especially when costly training procedures need to be avoided.

PRIMIS: Privacy-preserving medical image sharing via deep sparsifying transform learning with obfuscation

Objective:The primary objective of our study is to address the challenge of confidentially sharing medical images across different centers. This is often a critical necessity in both clinical and research environments, yet restrictions typically exist due to privacy concerns. Our aim is to design a privacy-preserving data-sharing mechanism that allows medical images to be stored as encoded and obfuscated representations in the public domain without revealing any useful or recoverable content from the images. In tandem, we aim to provide authorized users with compact private keys that could be used to reconstruct the corresponding images. Method:Our approach involves utilizing a neural auto-encoder. The convolutional filter outputs are passed through sparsifying transformations to produce multiple compact codes. Each code is responsible for reconstructing different attributes of the image. The key privacy-preserving element in this process is obfuscation through the use of specific pseudo-random noise. When applied to the codes, it becomes computationally infeasible for an attacker to guess the correct representation for all the codes, thereby preserving the privacy of the images. Results:The proposed framework was implemented and evaluated using chest X-ray images for different medical image analysis tasks, including classification, segmentation, and texture analysis. Additionally, we thoroughly assessed the robustness of our method against various attacks using both supervised and unsupervised algorithms. Conclusion:This study provides a novel, optimized, and privacy-assured data-sharing mechanism for medical images, enabling multi-party sharing in a secure manner. While we have demonstrated its effectiveness with chest X-ray images, the mechanism can be utilized in other medical images modalities as well.

Robust Data-Driven Automation Based on Relaxed Supervised Hashing With Self-Optimized Labels

Recently, the Visual Internet of Things (VIoT) has been widely used in data-driven automation, where VIoT devices are used to monitor environmental dynamics and to trigger corresponding actuators after examining event signatures (e.g., hashes). Nonetheless, VIoT devices may collect partially observed data, e.g., largely occluded images, which cause biased labeling and oversensitivity during modeling. Moreover, in typical methods, class labels are rigid and fixed categorical variables, where marginal space between different classes is fixed and equal. In fact, margins may be unequally spaced. Such nonflexibility in class labels inevitably causes fitting difficulty. In light of such, this study proposes relaxed robust supervised hashing (Relaxed RSH) for generating reliable signatures that can simultaneously conquer the above problems caused by incomplete data and rigid margins. To accommodate oversensitivity, this study proposes measuring hash learning loss by robust half quadratic (HQ) functions for modeling incomplete data. In the initial label matrix, slack variables are added to relax binary constraints. Such slack variables can be self-optimized during the learning process and can be used to automatically adjust margins between different classes. Decorrelation, balancing, and normalization constraints based on Relaxed RSH are also devised to provide discriminant and compact codes. Experimental results based on open datasets showed that the proposed method yielded higher mAP and F1 than the baselines. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This work is motivated by the problems caused by incomplete data and rigid class margins during event signature (e.g., hash) learning, where hashes are used to trigger automation systems. In existing hashing methods, incomplete data (e.g., continuous occlusion, missing values, and sample-specific outliers) along with rigid class margins cause fitting biases, thereby challenging data-driven automation. This study proposes Relaxed RSH by designing robust HQ loss, self-optimized label learning, and corresponding constraints for generating discriminant hashes. This prevents actuators from being mistriggered by corrupted data. Experiments were conducted on various data corruption. Future research will address the design for incremental/decentralized hash learning.

Mesoscale material modeling with memoryless isotropic point particles

There has been a proliferation of particle systems developed to model complex systems. These are attractive because they are mesh-free, avoiding issues associated with solver remeshing and convergence. They have however fragmented into niches, using increasingly complex particles that introduce internal degrees of freedom and external solver coupling. We show that, contrary to prior assumptions in the literature, memoryless isotropic point particles can model material properties including anisotropy, hysteresis, and failure solely through the statistics of their distributions. The resulting models offer compact code that is straightforward to accelerate and port, can span between micro- and macro-structure, require few parameters to set up a simulation, and unlike high-dimensional machine learning models they use a low-dimensional representation that can be efficiently learned. Rather than deriving them as approximations to either molecular dynamics or partial differential equations we investigate how these models can be found directly, and illustrate this with both qualitative comparisons of phenomenology and quantitative comparisons of predictions.

Deep Bayesian Quantization for Supervised Neuroimage Search.

Neuroimage retrieval plays a crucial role in providing physicians with access to previous similar cases, which is essential for case-based reasoning and evidence-based medicine. Due to low computation and storage costs, hashing-based search techniques have been widely adopted for establishing image retrieval systems. However, these methods often suffer from nonnegligible quantization loss, which can degrade the overall search performance. To address this issue, this paper presents a compact coding solution namely Deep Bayesian Quantization (DBQ), which focuses on deep compact quantization that can estimate continuous neuroimage representations and achieve superior performance over existing hashing solutions. Specifically, DBQ seamlessly combines the deep representation learning and the representation compact quantization within a novel Bayesian learning framework, where a proxy embedding-based likelihood function is developed to alleviate the sampling issue for traditional similarity supervision. Additionally, a Gaussian prior is employed to reduce the quantization losses. By utilizing pre-computed lookup tables, the proposed DBQ can enable efficient and effective similarity search. Extensive experiments conducted on 2, 008 structural MRI scans from three benchmark neuroimage datasets demonstrate that our method outperforms previous state-of-the-arts.

A novel homogenization method for periodic piezoelectric composites via diffused material interface

In this article, we propose a novel homogenization method for piezoelectric composites with periodic microstructures. In the proposed homogenization method, one can find the analytical expressions for the material properties of the composites in the representative volume element (RVE) via diffused material interface method. The availability of analytical expressions for the material properties in the entire domain of the RVE enables one to determine the effective homogenized material properties with the standard finite element method. As the proposed homogenization method regularizes the discontinuity in material properties across the interfaces, there is no need to explicitly track the material interfaces while implementing the finite element method for determining the effective material properties. Hence, one can implement the proposed homogenized method for any piezoelectric composites with complicated material interfaces and determine the effective material properties. In this study, we have considered the piezoelectric composites to be comprised of periodic microstructures where the RVE constitutes a matrix with an inclusion of a specific shape. We have carried out a study on the effect of the shape of inclusion viz. square-shaped, I-shaped, T-shaped, and plus-shaped, and the size of inclusion on the effective piezoelectric material properties. We have also studied the influence of shape and size of inclusion on the electro-mechanical response of a homogenized piezoelectric continuum. To implement the proposed homogenization method, we have used Gridap, an open-source finite element toolbox in Julia that provides very compact codes freely available to all the researchers and makes a third-party verification of the proposed method straightforward.

GraphIQA: Learning Distortion Graph Representations for Blind Image Quality Assessment

A good distortion representation is crucial for the success of deep blind image quality assessment (BIQA). However, most previous methods do not effectively model the relationship between distortions or the distribution of samples with same distortion type but different distortion levels. In this work, we start from the analysis of the relationship between perceptual image quality and distortion-related factors, such as distortion types and levels. Then, we propose a Distortion Graph Representation (DGR) learning framework for IQA, named GraphIQA, in which each distortion is represented as a graph, i.e., DGR. One can distinguish distortion types by learning the contrast relationship between these different DGRs, and infer the ranking distribution of samples from different levels in a DGR. Specifically, we develop two sub-networks to learn the DGRs: a) Type Discrimination Network (TDN) that aims to embed DGR into a compact code for better discriminating distortion types and learning the relationship between types; b) Fuzzy Prediction Network (FPN) that aims to extract the distributional characteristics of the samples in a DGR and predicts fuzzy degrees based on a Gaussian prior. Experiments show that our GraphIQA achieves the state-of-the-art performance on many benchmark datasets of both synthetic and authentic distortions. The code is available at http://staff.ustc.edu.cn/~chenzhibo/resources/2021/GraphIQA.html.

Multi-view clustering with orthogonal mapping and binary graph

The multi-view clustering problem has attracted considerable attention over recent years for the remarkable clustering performance due to exploiting complementary information from multiple views. Most existing related research work processes data in the decimal real value space that is not the most compatible space for computers. Binary code learning, also known as hashing technology, is well-known for fast Hamming distance computation, less storage requirement and accurate calculation results. The Hamming space is most enjoyed by computers because of binary/hash codes. Several studies combine multi-view clustering with binary code learning for improving clustering performance. However, there is much redundant information contained in the learned binary codes, which negatively affects the clustering performance, but these studies ignore eliminating redundant information for learning compact codes. In addition, they don’t give a unified (one-step) clustering framework with binary graph structure, which doesn’t lead to the optimal clustering result due to the information loss during the two-step process. In this paper, to cope with the two issues, we propose an orthogonal mapping binary graph method (OMBG) for the multi-view clustering problem, which makes the mapping matrix of every view orthogonalize for eliminating redundant information and embeds a binary graph structure into the unified binary multi-view clustering framework for extracting local geometric structure information of binary codes and achieving the optimal clustering result. Furthermore, we design an effective optimization algorithm based on alternating direction minimization to solve the model of OMBG. Extensive experiments performed on four frequently-used benchmark multi-view datasets illustrate the superiority of OMBG which is compared with some state-of-the-art clustering baselines.

Digital computing through randomness and order in neural networks

We propose that coding and decoding in the brain are achieved through digital computation using three principles: relative ordinal coding of inputs, random connections between neurons, and belief voting. Due to randomization and despite the coarseness of the relative codes, we show that these principles are sufficient for coding and decoding sequences with error-free reconstruction. In particular, the number of neurons needed grows linearly with the size of the input repertoire growing exponentially. We illustrate our model by reconstructing sequences with repertoires on the order of a billion items. From this, we derive the Shannon equations for the capacity limit to learn and transfer information in the neural population, which is then generalized to any type of neural network. Following the maximum entropy principle of efficient coding, we show that random connections serve to decorrelate redundant information in incoming signals, creating more compact codes for neurons and therefore, conveying a larger amount of information. Henceforth, despite the unreliability of the relative codes, few neurons become necessary to discriminate the original signal without error. Finally, we discuss the significance of this digital computation model regarding neurobiological findings in the brain and more generally with artificial intelligence algorithms, with a view toward a neural information theory and the design of digital neural networks.

Development of algorithms that simulate the solution of a problem by a person

Some algorithms, which are often based on the use of elements of higher mathematics, possessing high speed and compact coding in algorithmic languages, are poorly mastered by most students. It can be assumed that this is due to the difficulty of presenting the principles of their work in the form of human actions in ordinary situations. Thus, a certain contradiction arises between the way of solving the problem that a person resorts to without using a computer and the way we force our computer to solve this problem. Comparison of the process of explaining algorithms speaks in favor of algorithms imitating human thinking. The discussion of the advantages of the algorithms themselves is beyond the scope of this article and undoubtedly deserves a separate study. If artificial intelligence is created, then its creator or creators will certainly be ranked among the outstanding geniuses in the history of civilization, no matter what algorithms it uses. However, so far there is no one to solve problems for us and create algorithms, so we will use all available means and try to teach this to children.

Dual Distance Optimized Deep Quantization With Semantics-Preserving

Recently, quantization has been an effective technique for large-scale image retrieval, which can encode feature vectors into compact codes. However, it is still a great challenge to improve the discriminative capability of codewords while minimizing the quantization error. This letter proposes Dual Distance Optimized Deep Quantization (D <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ODQ) to deal with this issue, by minimizing the Euclidean distance between samples and codewords, and maximizing the minimum cosine distance between codewords. To generate the evenly distributed codebook, we find the general solution for the upper bound of the minimum cosine distance between codewords. Moreover, scaler constrained semantics-preserving loss is considered to avoid trivial quantization boundary, and ensure that a codeword can only quantize the features of one category. In contrast to state-of-the-art methods, our method has a better performance on three benchmark datasets.

Local symmetric directional pattern: A novel descriptor for extracting compact and distinctive features in face recognition

This paper presents a novel approach in feature extraction, Local Symmetric Directional Pattern (LSDP), for face recognition. The LSDP encodes the structure of facial textures based on gradient information in a simple and compact coding approach to produce more distinctive code in less time and memory than existing methods. We extract gradient information by convolving the face image with four symmetric compass masks to encode this information using directional numbers, which are related to directional information, and magnitudes of the two prominent edge responses. We also use a hybrid feature vector as a face descriptor obtained by reducing the dimensions of the LSDP feature map and classify them using the sparse representation-based classification (SRC) algorithm. Due to the high discrimination power of the extracted features, the construction of a dictionary based on the hybrid features leads to more sparse representation coefficients. As a result, it improves SRC performance in terms of recognition rate and computational speed. We perform extensive experiments to evaluate and compare the performance of our descriptor with other handcrafted descriptors and the deep learning feature under two different evaluation protocols (different dimensions of feature space and the different number of training samples) on different face databases, which have different variations of illumination, expression, and pose. Experimental results illustrate that our method achieves the highest recognition rate compared to other methods in both evaluation protocols. Especially under challenging conditions where the dimensions of the feature space or the number of training samples are low, LSDP demonstrates excellent performance.

Highly Information and Energy-efficient Monitoring Data Transmission in IoT Networks

The components of technologies of increasing information and energy efficiency of IoT monitoring networks subscriber systems with protected transfer functioning of reliable packets of information with the increased information capacity are described. In the places of formation of network flows, i.e. in the places of installation of object and on-board systems the realization of a complex of adaptive filtering algorithms, compression and protection of samples of monitoring signals and video data frames with the subsequent adaptive formation and transmission of highly informative code-signal sequences is offered. It is proposed a signal approach as a basis for inputting and compact coding of signals and video data frames of the reliable samples. The signal approach is proposed, according to which the amplitude-time or amplitude-number parameters of the most informative samples of signals and video signals are determined in the rate of input of monitoring data. These are extremes and points of inflection or points of curve movement change. The obtained data are subject to data compression with controlled information loss and lossless compression-protection. According to the proposed information technology for building effective IoT networks for crypto protection of monitoring data arrays by processors of object and on-board systems, the use of disposable ciphers is proposed, which are the rules and parameters for generating crypto-resistant pseudo-random data arrays of a certain length. These rules and parameters are known only to the subscriber-transmitter and the subscriber-receiver of information packets and are used by network subscribers in the process of data compression with losses and without losses, in the process of crypto-resistant and noise-resistant information packets forming of limited duration with increased information capacity. Energy-efficient data packet transmission is based on a significant reduction in the output streams of protected highly informative monitoring data packets and the implementation by object and on-board systems processors a set of algorithms for processing, encoding, encrypting and transmitting data minimized by computational complexity.

Deep Unsupervised Image Hashing by Maximizing Bit Entropy

Unsupervised hashing is important for indexing huge image or video collections without having expensive annotations available. Hashing aims to learn short binary codes for compact storage and efficient semantic retrieval. We propose an unsupervised deep hashing layer called Bi-Half Net that maximizes entropy of the binary codes. Entropy is maximal when both possible values of the bit are uniformly (half-half) distributed. To maximize bit entropy, we do not add a term to the loss function as this is difficult to optimize and tune. Instead, we design a new parameter-free network layer to explicitly force continuous image features to approximate the optimal half-half bit distribution. This layer is shown to minimize a penalized term of the Wasserstein distance between the learned continuous image features and the optimal half-half bit distribution. Experimental results on the image datasets FLICKR25K, NUS-WIDE, CIFAR-10, MS COCO, MNIST and the video datasets UCF-101 and HMDB-51 show that our approach leads to compact codes and compares favorably to the current state-of-the-art.

Weakly Supervised Deep Hyperspherical Quantization for Image Retrieval

Deep quantization methods have shown high efficiency on large-scale image retrieval. However, current models heavily rely on ground-truth information, hindering the application of quantization in label-hungry scenarios. A more realistic demand is to learn from inexhaustible uploaded images that are associated with informal tags provided by amateur users. Though such sketchy tags do not obviously reveal the labels, they actually contain useful semantic information for supervising deep quantization. To this end, we propose Weakly-Supervised Deep Hyperspherical Quantization (WSDHQ), which is the first work to learn deep quantization from weakly tagged images. Specifically, 1) we use word embeddings to represent the tags and enhance their semantic information based on a tag correlation graph. 2) To better preserve semantic information in quantization codes and reduce quantization error, we jointly learn semantics-preserving embeddings and supervised quantizer on hypersphere by employing a well-designed fusion layer and tailor-made loss functions. Extensive experiments show that WSDHQ can achieve state-of-art performance in weakly-supervised compact coding.

Motor synergies: Evidence for a novel motor signature in autism spectrum disorder

In autism spectrum disorder (ASD), socio-communicative impairments and stereotypical behaviours are paralleled by sensorimotor deficits. Individuals with ASD show an altered selection of motor parameters, resulting in clumsy and fragmented actions. Here, we investigated inter-joint coordination and motor synergies as a potential substrate of motor control problems in ASD. Synergies enable co-controlling redundant motor degrees of freedom (DoF, e.g. joint angles, muscles) by mapping behavioural goals into a flexible and low-dimensional set of variables. This mechanism simplifies motor control and helps to find unambiguous solutions for motor tasks. In a reaching-grasping paradigm, children with ASD showed reduced coupling between DoF, which correlated with socio-communicative symptoms severity. Impaired synergies may help to frame well-established motor problems in ASD, including impaired motor sequencing and abnormal trial-to-trial motor variability. On the other hand, synergies also provide an effective and compact coding system of observed actions. Impaired synergies may thus jeopardize motor interaction by initiating bottom-up cascade effects, leading to pervasive impairments of social behaviour. Finally, we trained an automatic classification algorithm to distinguish between ASD and typically developing (TD) participants based on reaching-grasping kinematics. Classification accuracy reached up to 0.947. This result corroborates and expands previous accounts claiming that motor-based early recognition is feasible and effective in ASD.

Deep supervised hashing using quadratic spherical mutual information for efficient image retrieval

Several deep supervised hashing techniques have been proposed to allow for extracting compact and efficient neural network representations for various tasks. However, many deep supervised hashing techniques ignore several information-theoretic aspects of the process of information retrieval, often leading to sub-optimal results. In this paper, we propose an efficient deep supervised hashing algorithm that optimizes the learned compact codes using an information-theoretic measure, the Quadratic Mutual Information (QMI). The proposed method is adapted to the needs of efficient image hashing and information retrieval leading to a novel information-theoretic measure, the Quadratic Spherical Mutual Information (QSMI). Apart from demonstrating the effectiveness of the proposed method under different scenarios and outperforming existing state-of-the-art image hashing techniques, this paper provides a structured way to model the process of information retrieval and develop novel methods adapted to the needs of different applications.

A complexity reduced and reliable integrity protection for large relational data over clouds

At present, governments and private business operations are highly dependent on relational data applications such as bank accounts, citizen registration, etc. These relational data dependent operations require reliable integrity protection while utilising the cloud computing storage infrastructure. Identification and recovery of stolen bits are a major assistance to the reliable integrity protection services for the sensitive relational data applications. To deal with the problems of detecting and recovering tampering in large relational data at minimum computational complexity, in this paper, N8WA (briefed in Section 2.1) coding-based scheme is presented. Overall the scheme is comprised of two cross functional modules. The first module is labelled as compact code generation using N8WA coding and code registration at registration module (RM). In the second module which is called accurate locating/restoring tampering, utilising the mismatching of different compact codes based on N8WA from RM, the major/minor tampered data is accurately located and restored. Investigational outcome indicates that the scheme ensures the computational complexity of O(n2) while minimum to maximum alterations is accurately localised and restored successfully.