Proposed Compression Method Research Articles

Literature Study of Learning-Based Video Compression

Developments in telecommunications technology today, such as cellular with the fifth generation (5G), the development of IoT prototypes, and the migration of analog TV to digital TV starting in 2022. The development of various research using machine learning. The problem with video format information is that the video file size is quite large, so the transmission process requires a large bandwidth. In addition, sharing services such as Video on Demand (VoD) and Video Broadcasting are sensitive to delay. In comparison, the transmission media has limited capacity, such as terrestrial TV, Ethernet/Fast Ethernet, and wireless cellular data such as 2G, 3G HSPA, 4G, etc. Based on reports from Cisco, the development of internet users has increased by 10% per year, with 80% of total traffic using video. Developments in various video compression standards, such as the most recent H.264 and H.265, produce high-quality, low-bitrate video. Much research has been carried out with various proposed compression methods based on machine learning. Either uses singular block learning based or end-to-end. This research focuses on the literature study of video compression with machine learning.

An Area-Efficient Large Integer NTT-Multiplier Using Discrete Twiddle Factor Approach

Number theoretic transform (NTT) method shows great advantages in speed and efficiency for applications such as homomorphic encryption. However, the twiddle factor data consumes a lot of memories. In this brief, we propose a novel data compression method, together with the corresponding data storage scheme and addressing algorithm. Furthermore, we design a 768k-bit multiplier with a full pipeline structure. Our proposed compression method has achieved a compression rate of 98.8% for the twiddle factor data. Compared with the state-of-the-art FPGA implementations, our design shows up to 44.2% improvement in terms of area-efficiency.

Model compression optimized neural network controller for nonlinear systems

Neural network-based controllers are widely used within the domain of robotic control systems. A network controller with more neurons typically achieves better performance, but an excessive number of neurons may make the model computationally intensive, resulting in slow dynamic responses in real-world environments. This paper reports a network compression method that integrates knowledge distillation technology for the development of concise neural network-based controllers to achieve a balance between the control performance and computational costs. The method first trains a full-size teacher model, which is then pruned, leading to a concise network with a minimum compromise of performance. From in this study, the resulting concise network is considered to be the prototype of a student model, which is further trained by a knowledge distillation process. The proposed compression method was applied to three classical networks, and the resultant compact controllers were tested on a robot manipulator for efficacy and potential demonstration. The experimental results from a comparative study confirm that the student models with fewer neurons resulting from the proposed model compression approach can achieve similar performance to that of the teacher models for intelligent dynamic control but with faster convergence speed.

Versatile Video Coding-Based Coding Tree Unit Level Image Compression With Dual Quantization Parameters for Hybrid Vision

Image analysis based on machine vision is hugely manipulated in the smart industry. Good-quality images are required for outstanding machine analysis results, but handling high-definition images could be problematic in a constrained environment such as a low-bandwidth network or low-capacity storage. Lowering the image resolution might be a straightforward solution for reducing image data, but it would occasion much information loss, leading to the deterioration of machine vision. Moreover, human supervision could be necessary for a contingency that machine vision cannot control.Therefore, an innovative image compression method considering machine and human vision is required; more compression efficiency than the state-of-the-art codec, praiseworthy machine vision performance, and human-recognizable quality. In this paper, we propose Versatile video coding(VVC) based image compression for hybrid vision, i.e., machine vision and human vision. Our work provides a coding tree unit(CTU) level image compression with dual quantization parameters (QPs) according to the quantization parameter map and the saliency extracted by the object detection network; in the salient region, the proposed method maintains high quality with low QP but degrades the quality with high QP in the non-salient region.Compared with VVC, the proposed compression method achieves a bitrate reduction of up to 25.5% in machine vision tasks, proving more compression efficiency and still admirable machine vision performance. From the perspective of human vision, the proposed method provides human-perceptible image quality, preserving acceptable objective quality values.

Silicone elastomers deformation peculiarities under compression

This paper deals with the polymer composite material unique properties on self-restoration after ultimate compressive loads application. Material is based on polydimethylsiloxane elastomeric matrix with carbonyl iron particles isotropic fillers in concentration 75% wt. This composite mechanical characteristic ranges were studied in comparison with the characteristics of conventional polymer based on silicone elastomer without the fillers. The investigated material, the so-called magneto-elastic, demonstrated ability to self-restore its shape after extreme loads removal is a promising direction for the revealed unique property use and will probably be demanded for application in number fields of science and technology. The tests were carried out on an electromechanical testing machine. These materials investigating various compression modes results summarized as: the strain diagrams changes at different load ranges were plotted, the characteristics changes in their deformation properties were analyzed, the compression deformation linear and nonlinear characteristics limits were determined. The proposed compression method in various modes made it possible to determine the load changes ranges causing both irreversible changes in the conventional polymer based on silicone elastomer and its further destruction and the unique capability for self-recovery in a composite material, the so-called magneto-elastic, after similar types ultimate loads application to it.

BΔ-NIS: Performance analysis of an efficient data compression technique for on-chip communication network

Nowadays, system-on-chip (SoC) integrates more processing cores and IPs with network-on-chip (NoC) for satisfying today's demands, including low power consumption and high performance, without affecting the elasticity and scalability. When the on-chip processing elements (PEs) are increased, burst traffic will be generated from memory and cache accesses. As a result, the packet transmission latency is increased, reducing the system's speed performance. This paper proposes a new hybrid Base delta Neighbourhood Indexing Sequence (BΔ-NIS) based compression method for compressing the network traffic. The previous solutions for data compression often contain incompressible packets for certain data patterns. The proposed method uses Neighbourhood Indexing Sequence (NIS) algorithm to determine the optimal codeword for uncompressible chunks of delta compression approach using zeros and ones based traversals. Because of the dynamic nature of BΔ-NIS, it is highly effective and compresses all chunks in the packets efficiently. Moreover, the concept of dynamic header flit extension is introduced in the proposed compression method, which provide unlimited scalability. The proposed BΔ-NIS algorithm is robust to packet losses and leads to significant power saving. The simulation results show that the proposed BΔ-NIS reduces the hardware overhead and overtakes the existing data compression techniques by attaining a mean compression ratio of 0.73. In addition, BΔ-NIS accomplishes 51% latency reduction compared to the baseline router (without any compression). It also minimizes the power consumption of the de/compression modules by 70.14% and 20.8% as compared to the NoΔ and FlitZip respectively.

The Lossless Adaptive Binomial Data Compression Method

In this paper, we propose a new method for the binomial adaptive compression of binary sequences of finite length without loss of information. The advantage of the proposed binomial adaptive compression method compared with the binomial compression method previously developed by the authors is an increase in the compression rate. This speed is accompanied in the method by the appearance of a new quality—noise immunity of compression. The novelty of the proposed method, which makes it possible to achieve these positive results, is manifested in the adaptation of the compression ratio of compressible sequences to the required time, which is carried out by dividing the initial set of binary sequences into compressible and incompressible sequences. The method is based on the theorem proved by the authors on the decomposition of a stationary Bernoulli source of information into the combinatorial and probabilistic source. The last of them is the source of the number of units. It acquires an entropy close to zero and practically does not affect the compression ratio at considerable lengths of binary sequences. Therefore, for the proposed compression method, a combinatorial source generating equiprobable sequences is paramount since it does not require a set of statistical data and is implemented by numerical coding methods. As one of these methods, we choose a technique that uses binomial numbers based on the developed binomial number system. The corresponding compression procedure consists of three steps. The first is the transformation of the compressible sequence into an equilibrium combination, the second is its transformation into a binomial number, and the third is the transformation of a binomial number into a binary number. The restoration of the compressed sequence occurs in reverse order. In terms of the degree of compression and universalization, the method is similar to statistical methods of compression. The proposed method is convenient for hardware implementation using noise-immune binomial circuits. It also enables a potential opportunity to build effective systems for protecting information from unauthorized access.

Hybrid Model Compression for Multi-Task Network

In recent years, deep neural networks have made significant achievements in various fields in science and engineering. To achieve outstanding performance, neural network models with very deep structures and an extensive number of parameters are usually used, leading to huge memory storage and computational resources requirements of the hardware systems, and thus preventing them from being deployed in devices with hardware limits. In particular, such problem will be further amplified when multiple neural network models for several tasks are to be integrated into one single platform at the same time. In this paper, a hybrid joint-network optimization model is proposed to solve this multi-task multi-model compression problem. The idea of hard parameter sharing in multi-task learning is adopted. Specifically, we turn the sub-network of the same structure of multiple models into a hybrid network model, in which the same parameters are shared for all the tasks. In addition, a unique sparse matrix for each task is added to each task independently. Experimental results show that the proposed compression methods can not only achieve a high compression ratio but also get performance improvement for some tasks.

Video Image Processing Method Based on Cloud Platform Massive Data and Virtual Reality

In order to solve the practical problem that the massive video data generated by monitoring equipment cannot be processed efficiently temporarily, this paper proposes a framework for face recognition of massive video data based on distributed environment. Combined with the application features of face recognition, this method designs a strategy for fast reading of massive video data and optimizes the feature data obtained by the cloud platform, so as to speed up the retrieval speed of face features. The results are as follows: the compression rate of the proposed compression method is higher than that of the traditional matrix triple and binary methods, which is increased by about 65%; the data optimization method in this paper greatly reduces the amount of feature data, which is 7.08 times less than that in the nonoptimization state. At the same time, the process of face recognition is reduced from 12.6 seconds to 2.73 seconds, and the time of feature decompression is only 0.75 seconds more than the original; the experiment shows that it takes 10180 seconds for the system to process 200 GB pictures with 9 computing nodes, and the total running time of the system is 4737 seconds longer than that of a single node, accounting for about 5.45% of the total time of a single node system. At the same time, the experimental data show that the system is 8.53 times faster than that of a single node with 9 computing nodes. It is proved that this framework has certain research significance in dealing with massive unstructured data. It not only provides theoretical reference value for the research of massive video processing but also makes a contribution to the actual industry.

MGARD+: Optimizing Multilevel Methods for Error-Bounded Scientific Data Reduction

Nowadays, data reduction is becoming increasingly important in dealing with the large amounts of scientific data. Existing multilevel compression algorithms offer a promising way to manage scientific data at scale, but may suffer from relatively low performance and reduction quality. In this paper, we propose MGARD+, a multilevel data reduction and refactoring framework drawing on previous multilevel methods, to achieve high-performance data decomposition and high-quality error-bounded lossy compression. Our contributions are four-fold: 1) We propose to leverage a level-wise coefficient quantization method, which uses different error tolerances to quantize the multilevel coefficients. 2) We propose an adaptive decomposition method which treats the multilevel decomposition as a preconditioner and terminates the decomposition process at an appropriate level. 3) We leverage a set of algorithmic optimization strategies to significantly improve the performance of multilevel decomposition/recomposition. 4) We evaluate our proposed method using four real-world scientific datasets and compare with several state-of-the-art lossy compressors. Experiments demonstrate that our optimizations improve the decomposition/recomposition performance of the existing multilevel method by up to <inline-formula><tex-math notation="LaTeX">$70 \times$</tex-math></inline-formula> , and the proposed compression method can improve compression ratio by up to <inline-formula><tex-math notation="LaTeX">$2 \times$</tex-math></inline-formula> compared with other state-of-the-art error-bounded lossy compressors under the same level of data distortion.

Three-Dimensional Adaptive Image Compression Concept for Medical Imaging: Application to Computed Tomography Angiography for Peripheral Arteries.

Advances in computed tomography (CT) have resulted in a substantial increase in the size of datasets. We built a new concept of medical image compression that provides the best compromise between compression rate and image quality. The method is based on multiple contexts and regions-of-interest (ROI) defined according to the degree of clinical interest. High priority areas (primary ROIs) are assigned a lossless compression. Other areas (secondary ROIs and background) are compressed with moderate or heavy losses. The method is applied to a whole dataset of CT angiography (CTA) of the lower extremity vasculature. It is compared to standard lossy compression techniques in terms of quantitative and qualitative image quality. It is also compared to standard lossless compression techniques in terms of image size reduction and compression ratio. The proposed compression method met quantitative criteria for high-quality encoding. It obtained the highest qualitative image quality rating score, with a statistically significant difference compared to other methods. The average compressed image size was up to 61% lower compared to standard compression techniques, with a 9:1 compression ratio compared with original non-compressed images. Our new adaptive 3D compression method for CT images can save data storage space while preserving clinically relevant information.

A Performance Analysis of Edge Computing Compression Methods for Environmental Monitoring Nodes with LoRaWAN Communications

Data compression methods applied with wavelet transforms, redundant point removal, or neural networks are potential edge computing solutions whose main advantages are minimization of the amount of data transmitted in wireless sensor networks and the reduction of energy requirements and consumption. Implemented as central computational unit, the microcontroller used in such systems compresses the data, specifically temperature data, acquired from environmental monitoring. The proposed compression methods can achieve up to 93% data compression at a mean squared error value of approximately 0.1. The method was designed to satisfy the requirements of the LoRaWAN protocol and the size of the data payload.

Image Compression Using Stochastic-AFD Based Multisignal Sparse Representation.

Adaptive Fourier decomposition (AFD) is a newly developed signal processing tool that can adaptively decompose any single signal using a Szegö kernel dictionary. To process multiple signals, a novel stochastic-AFD (SAFD) theory was recently proposed. The innovation of this study is twofold. First, a SAFD-based general multi-signal sparse representation learning algorithm is designed and implemented for the first time in the literature, which can be used in many signal and image processing areas. Second, a novel SAFD based image compression framework is proposed. The algorithm design and implementation of the SAFD theory and image compression methods are presented in detail. The proposed compression methods are compared with 13 other state-of-the-art compression methods, including JPEG, JPEG2000, BPG, and other popular deep learning-based methods. The experimental results show that our methods achieve the best balanced performance. The proposed methods are based on single image adaptive sparse representation learning, and they require no pre-training. In addition, the decompression quality or compression efficiency can be easily adjusted by a single parameter, that is, the decomposition level. Our method is supported by a solid mathematical foundation, which has the potential to become a new core technology in image compression.

SAR Image Compression Using Discretized Gaussian Adaptive Model and Generalized Subtractive Normalization

Synthetic aperture radar (SAR) image compression plays an important role in the manipulation of images. However, existing optical compression methods cannot properly handle SAR compression due to the absence of feature learning and representation for SAR images. In this letter, we propose an end-to-end trainable model to effectively fit the feature distribution and reduce information dependencies toward SAR image compression. To better parameterize the distribution of latent codes, a discretized Gaussian adaptive model is designed to achieve a flexible entropy process. To further remove the remaining redundancies, generalized subtractive normalization is introduced to reduce the statistical dependencies in SAR images. Extensive experiments show that the proposed compression method outperforms the traditional compression methods and learning-based algorithms on both the ICEYE and Sandia datasets.

Rolling bearing fault diagnosis with compressed signals based on hybrid compressive sensing

In recent years, many studies have been conducted on remote machine condition monitoring and on-board bearing fault diagnosis. To accurately identify the fault pattern, large amounts of vibration data must be sampled and saved. However, because the bandwidth of the wireless communication is limited, the volume of transmitted data cannot be too large. Therefore, raw data are usually compressed before transmission. All of the previous compression methods use a sample-then-compress framework; however, in this work, we introduce a compression method based on a Compressive Sensing that can simultaneously collect and compress raw data. Additionally, a hybrid measurement matrix is designed for Compressive Sensing and used to compress the data and make it easy to conduct fault diagnosis in the compression domain. This process can significantly reduce the computational complexity of on-board fault diagnosis. The experimental results demonstrate that the proposed compression method is easy to use and the reconstruction process can recover the original signal perfectly. Above all, the compressed signal can preserve the time series information of the original signal, and in contrast to fault diagnosis using the original signal, the accuracy of the fault diagnosis based on conventional methods in the compression domain is not significantly decreased.

3D image compression during ultrasound phased diagnostics based on wavelet subband coding planar scans

The proposed compression method is based on the application of a two-dimensional discrete fast wavelet transform (FWT) to planar scans of 3D ultrasound images in order to simultaneously reduce redundancy and suppress speckle at a fixed quota of bits. The main idea of the method is to fuse three rules for threshold processing the wavelet coefficients of the scans, uniform and non-uniform quantizers, and bit quota distributions over subbands of the scan FWT based on the proposed cost function. The simulation results have shown that at the encoding rate of up to 1 bit/pixel, the quantity of artefacts were decreased up to 5-7 % of the original quantity under a signal-to-speckle ratio more than 16 dB, and the structural similarity index (SSIM) increased to 0.94-0.97 for defects of rectangular, triangular and oval shapes. The paper also presents the results proving the effectiveness of the proposed method in comparison with some variants of the solution according to the scheme “pre-filtering + codec”.

Unsupervised Channel Compression Methods in Motor Prostheses Design.

The development of high performance brain machine interfaces (BMIs) requires scaling recording channel count to enable simultaneous recording from large populations of neurons. Unfortunately, proposed implantable neural interfaces have power requirements that scale linearly with channel count. To facilitate the design of interfaces with reduced power requirements, we propose and evaluate an unsupervised-learning-based compressed sensing strategy. This strategy suggests novel neural interface architectures which compress neural data by methodically combining channels of spiking activity. We develop an entropy-based compression strategy that models the population of neurons as being generated from a lower dimensional set of latent variables and aims to minimize the loss of information in the latent variables due to compression. We evaluate compressed features by inferring the latent variables from these features and measuring the accuracy with which the activity of held out neurons and arm movements can be estimated. We apply these methods to different cortical regions (PMd and M1) and compare the proposed compression methods to a random projections strategy often employed for compressed sensing and to a supervised regression based channel dropping strategy traditionally applied in BMI applications.

Recursive Prediction for Lossless Image Compression

This paper introduced an algorithm for lossless image compression to compress natural and medical images. It is based on utilizing various casual fixed predictors of one or two dimension to get rid of the correlation or spatial redundancy embedded between image pixel values then a recursive polynomial model of a linear base is used.&#x0D; The experimental results of the proposed compression method are promising in terms of preserving the details and the quality of the reconstructed images as well improving the compression ratio as compared with the extracted results of a traditional linear predicting coding system.

FCNN: An Efficient Intrusion Detection Method Based on Raw Network Traffic

When traditional machine learning methods are applied to network intrusion detection, they need to rely on expert knowledge to extract feature vectors in advance, which incurs lack of flexibility and versatility. Recently, deep learning methods have shown superior performance compared with traditional machine learning methods. Deep learning methods can learn the raw data directly, but they are faced with expensive computing cost. To solve this problem, a preprocessing method based on multipacket input unit and compression is proposed, which takes m data packets as the input unit to maximize the retention of information and greatly compresses the raw traffic to shorten the data learning and training time. In our proposed method, the CNN network structure is optimized and the weights of some convolution layers are assigned directly by using the Gabor filter. Experimental results on the benchmark data set show that compared with the existing models, the proposed method improves the detection accuracy by 2.49% and reduces the training time by 62.1%. In addition, the experiments show that the proposed compression method has obvious advantages in detection accuracy and computational efficiency compared with the existing compression methods.

Efficient multichannel EEG compression by optimal tensor truncation

Electroencephalography (EEG) signal is a crucial biomedical signal that helps assess the psychological, neural, and cognitive conditions of the human brain. However, the huge size of modern multi-channel EEG signals sampled at a high rate create difficulties in storage, transmission, and processing. Efficient compression methods obtain a compact form of the signal and mitigate these issues. In this paper, we propose an optimal tensor truncation method for performing compression of the data. In our proposed work, we first reshape the multi-channel EEG signal as a tensor and initially identify the optimum size of the compressed tensor. In the next stage, we perform tensor decomposition and obtain a core tensor of significantly smaller size to represent the information contained in the tensor. The core tensor obtained by the process captures the details of the original signal and leads to high compression ratio. We reconstruct the original tensor by transforming the core tensor with the transposed factor matrices on the decoder side. We have experimented with the proposed compression method on several online available datasets using several performance measures. From the experiments mentioned above, we conclude that we obtain better compression performance compared to the state of the art approaches.