Earticle

Genetic algorithm based predictor for lossless image compression is propsed. We describe a genetic algorithm to learn predictive model for lossless image compression. The error image can be further compressed using entropy coding such as Huffman coding or arithmetic coding. We show that the proposed algorithm can be feasible to lossless image compression algorithm.

모델 인버전은 원본 학습 데이터 없이, 사전 학습된 모델로부터 반복적인 최적화를 통해 합성 입력을 복원하는 데이 터 없는 학습에서 널리 사용되는 기법이다. 그러나 최신 비전 트랜스포머에 이를 적용할 경우, 고비용의 셀프 어텐 션 메커니즘으로 인해 큰 계산적 부담이 발생하게 된다. 이를 중요하지 않은 패치들을 모두 제거함으로써 효율성을 향상시키는 희소 모델 인버전이 제안되었다. 하지만 데이터가 없는 상황에서 검증 데이터의 부재로 인한 학습 불안 정성의 증폭은 여전히 해결해야 할 문제로 남아 있다. 검증 데이터가 없는 환경에서는 모델의 정확도가 불확실해지 고, 변동성이 커지므로, 모델의 강건성 향상이 필수적이다. 본 논문에서는 데이터 없는 환경에서 생성되는 이미지의 품질과 다양성을 일관되게 유지하여, 정확도에 대한 표준편차를 낮추고 강건성을 향상시키는 방법을 제안한다. 제안 한 Adaptive AEM은 패치 제거 이후의 중요도를 재조정해 엔트로피 최소화를 촉진시킨다. 실험 결과, 제안한 방법 으로 생성된 이미지를 사용하면 이전 방법론에 비해 데이터 없는 양자화에서는 최대 72%, 데이터 없는 지식 증류 에서는 최대 49%까지 정확도의 표준편차를 줄여 모델을 강건하게 만들 수 있음을 입증한다.

Model inversion is a widely used technique in data-free learning, where synthetic inputs are reconstructed from a pretrained model through iterative optimization without access to the original training data. However, when applied to modern Vision Transformers, the high computational cost of the self-attention mechanism poses a significant challenge. Sparse Model Inversion (SMI) has been proposed to improve efficiency by removing non-essential patches. Nevertheless, in the absence of real validation data, the instability of training remains an unresolved issue, as model accuracy becomes uncertain and exhibits high variance. To address this, we propose a method that consistently preserves the quality and diversity of generated images in data-free environments, thereby reducing the standard deviation of accuracy and enhancing model robustness. The proposed Adaptive AEM readjusts the importance after patch removal to promote entropy minimization. Experimental results demonstrate that using images generated by our method reduces the standard deviation of accuracy by up to 72% in data-free quantization and up to 49% in data-free knowledge distillation, compared to previous approaches, leading to significantly more robust models.

기종점 데이터는 수요 분석 및 서비스 설계를 위해서 대중교통, 도로운영 등 다양한 분야에 서 저장 및 활용되고 있다. 최근 빅데이터의 활용성이 증대되면서 기종점 데이터의 분석 및 활용에 대한 수요도 함께 증가하고 있다. 기존의 일반적인 교통 정보 데이터가 수집장비 수(n) 에 비례하여 데이터양이 증가(a·n)하는 것과는 다르게, 기종점 데이터는 수집지점 수(n)의 증 가에 따라 수집 데이터의 양이 기하급수적으로 증가(a·n2)하는 경향이 있다. 이로 인하여 기종 점 데이터를 원시 데이터의 형태로 장기간 저장하고 빅데이터 분석에 활용하는 것은 대용량의 저장 공간이 필요하다는 것을 고려할 때 실용적 대안으로 여겨지지 않고 있다. 이와 함께 기종 점 데이터는 0~10 사이의 작은 수요 부분에 패턴화된 형태와 무작위 적인 형태의 데이터가 섞여있어 작은 수요가 그룹화되어 발생하는 주요 패턴을 추출하기에 어려움이 있다. 이러한 기종점 데이터의 저장용량의 한계와 패턴화 분석의 한계를 극복하고자 본 연구에서는 주성분 분석을 활용한 대중교통 기종점 데이터의 압축 및 분석 방법을 제안하였다. 본 연구에서는 서 울시와 세종시의 대중교통 이용 데이터를 활용하여 모빌리티 데이터를 분석하고, 모빌리티 기 종점 데이터에 포함된 무작위 성향이 높은 데이터를 제거하기 위해 주성분분석 기반의 데이터 압축 및 복원에 관한 연구를 수행하였다. 주성분분석으로 분해된 기종점 데이터와 원데이터를 비교하여 주요한 수요 패턴을 찾고 이를 통해 압축률과 복원율을 높일 수 있는 주성분 범위를 제안하였다. 본 연구에서 분석한 결과, 서울시 기준 1~80, 세종시 기준 1~60까지의 주성분을 사용할 경우 주요 이동 데이터의 손실 없이 기종점 데이터에 포함되어있는 노이즈를 제거하고 데이터를 압축 및 복원이 가능하였다.

Origin-destination data have been collected and utilized for demand analysis and service design in various fields such as public transportation and traffic operation. As the utilization of big data becomes important, there are increasing needs to store raw origin-destination data for big data analysis. However, it is not practical to store and analyze the raw data for a long period of time since the size of the data increases by the power of the number of the collection points. To overcome this storage limitation and long-period pattern analysis, this study proposes a methodology for compression and origin-destination data analysis with the compressed data. The proposed methodology is applied to public transit data of Sejong and Seoul. We first measure the reconstruction error and the data size for each truncated matrix. Then, to determine a range of principal components for removing random data, we measure the level of the regularity based on covariance coefficients of the demand data reconstructed with each range of principal components. Based on the distribution of the covariance coefficients, we found the range of principal components that covers the regular demand. The ranges are determined as 1~60 and 1~80 for Sejong and Seoul respectively.

In view of the valuable medical resources at home and abroad, especially considering the fact that children's medical resources can't meet the needs, the author designed the mobile medical integration system with the help of mobile Internet platform and cloud computing platform. The system is divided into two parts of the mobile terminal and cloud computing platform, which are used respectively by the guardian and the doctor. The health monitoring terminal designed for children are wearable watches whileAPP is developed for the guardian and the doctors. Cloud platform designed a platform for data storage, message processing, functional applications and other modules, forming a “cloud+client” service model. This system makes the children's disease prevention, emergency treatment and medical treatment behavior become more convenient and fast, protects the healthy growth of children, provides reference and solution for the future development of medical care.

Dam deformation monitoring is quite important. Monitoring of dam deformation is important for ensuring people's normal lives and reducing losses. Dam deformation often experiences a relatively long time. The deformation rate is relatively small at the beginning time, and it is large at the later stage. Monitoring at the early stage of dam deformation is of great significance. The wireless sensor network has the characteristics of not being affected by the terrain, convenient arrangement and so on, which can be used to monitor the deformation of the dam. In order to improve the transmission efficiency of the wireless sensor network, data compression is helpful for the energy savings and transmission rate improving. In this paper, a dam monitoring system based on wireless sensor networks is developed. In this paper, a dam monitoring system based on wireless sensor network and GPS multi antenna is developed. In the network monitoring system and data acquisition using wireless sensor module, data transmission using data compression technology and data cache technology. At the end of the monitoring terminal, desktop cloud technology is used to process the data collected and show the real time dam deformation condition. This method can also be used in other object motion or deformation with large range and other monitoring field to improve the monitoring efficiency.

Deformation monitoring for dam is quite important and it can be directly used to reflect the running status to keep the operation safely. In order to meet the demand for monitoring the dam, global position system (GPS) is used to develop a monitoring system with high efficiency. In this paper, we developed a monitoring system for deformation monitoring for the dam with wireless data transmission. In order to reduce the energy consumption, the regulation of its working condition has been redesigned according to the monitoring object's own characteristics. At the same time, the data compression is also used in the GPS data processing to reduce the energy consumption. The compression method is divided into two steps. In the preprocessing, data amount can be removed about half of the amount, and Hoffman coding compression rate and the style conversion compression ratio are relatively small. The total compression rate is more than 60%. Compared with the existing compression algorithms, the new developed method has higher compression rate. When the data collection rate is higher, the compression will be higher, and the reason may be that the time information and position information are compressed when the data collection rate is higher. This method can be used to process the data effectively and improve the data transmitting efficiency.

Sensor data exhibit strong correlation in both space and time. Many algorithms have been proposed to utilize these characteristics. However, each sensor just utilizes neighboring information, because its communication range is restrained. Information that includes the distribution and characteristics of whole sensor data provides other opportunities to enhance the compression technique. In this paper, we propose an orthogonal approach for compressing sensor readings based on a novel feedback technique. That is, the base station or a super node generates Huffman code for the compression of sensor data and broadcasts it into sensor networks as Huffman code. All sensor nodes that have received the information compress their sensor data and transmit them to the base station. We call this approach as feedback-diffusion and this modified Huffman coding as sHuffman coding. In order to show the superiority of our approach, we compare it with the existing data compression algorithms in terms of the lifetime of the sensor network. As a result, our experimental results show that the whole network lifetime was prolonged by about 30%.

The Intensive Care Unit (ICU) bedside monitors present the medical staff with large amounts of continuous data which can create a number of challenges. If the data is transmitted as part of a telemedicine system then the large volume of data can put pressure on bandwidth and affect the quality of service of the network. Another challenge is that the large volume of data has to be interpreted by medical staff to make a patient state assessment. In this paper we propose a time series analysis technique called data wavelets to derive trends in the data – this acts as a form of data compression for telemedicine and improves the quality of service of a network and also facilitates clinical decision support in the form of qualitative reasoning for patient state assessment. Our approach has been successfully applied to cardiovascular data from a neonatal ICU.

Broadband is the common form of telecommunication used for Intensive Care Unit (ICU) telemedicine. However, in rural areas, bandwidth demand can easily outstrip the revenue realizable that is needed to pay for the network infrastructure investment so lower bandwidth is normal. A consequence of restricted bandwidth on access pipes is service contention at the customer site. To address these challenges we need to consider Quality of Service issues before we can successfully deploy a successful ICU telemedicine system. Quality of Service refers to the set of technologies and techniques for managing network traffic with the goal of providing a certain level of performance to a data flow in a network. In this paper we will discuss how the use of data wavelets as a form of data compression of ICU data makes for better use of broadband in rural areas and, in turn, improves Quality of Service in telemedicine.

In this paper, it is proposed to implement the image compression, encryption and ECG data compression using by binary description of Discrete Cosine Transform (DCT), Binary Haar Transform and Discrete Hartley Transform (DHT). In this Binary Discrete Cosine Transform (Binary DCT), Binary Haar Transform and Binary Discrete Hartley Transforms (Binary DHT) are developed using the Walsh Hadamard transform (WHT). The resulting transform nearly exact the underlying transform very well, while maintaining all the advantages and properties of WHT. The Binary DCT is a well known sequency ordered Walsh Hadamard Transform (WHT), where as the Binary DHT can be considered as a new Hartley ordered WHT. Specifically, the properties of the proposed Hartley ordering are discussed and a shift copy scheme is proposed for a simple and direct generation of the Hartley ordering functions.

A new distributed algorithm of data ccompression based on hierarchical cluster model for sensor networks s proposed, the basic ideas of which are as follows, firstly the whole sensor network is mapped into a kind f hierarchical clusters model, and then different wavelet transform models are used to commit data ompression in inner and super clusters respectively, according to the relative regularity of sensor nodes eployed in the inner clusters, and the relative irregularity of sensor nodes deployed in super cluster. heoretical analyses and simulation results show that, the above new methods have good performance of pproximation, and can compress data and reduce the amount of data efficiently. So, it can prolong the ifetime of the whole sensor network to a greater degree.

In this paper, a new ECG compression method called Wavelet Threshold Based Immune Algorithm (WTBIA) is proposed. This method based on finding the best threshold level for each wavelet subband using Immune Algorithm (IA). The WTBIA algorithm consists of three main steps: 1) Applying 1-D Discrete Wavelet Transform (DWT) on ECG signal; 2) Thresholding of wavelet coefficients in each subband; and 3) Minimization of the Percent Root mean square Difference (PRD) and maximization of the Compression Ratio (CR) using IA. The main advantage of this method is finding the best threshold level for each subband based on the required CR and PRD. The compression algorithm was implemented and tested upon records selected from the MIT-BIH arrhythmia database [6] using different wavelets such as Haar, Daubechies, Coiflet, Symlet and Biorthogonal. Simulation results show that the proposed algorithm leads to high CR associated with low distortion level relative to previously reported compression algorithms.

As the increasing usage of data generating devices like cameras, mobile phones, Auto-ID technologies, and so on, huge amount of data are created. Data compression is very important. In order to use the suitable compression methods on reducing the data volume, this paper uses iceberg-cubes to compress the data based on the entropy query mechanism. Using the definition of iceberg-cubes, this paper uses the entropy query principle for getting the key value from the original datasets. The iceberg-cubes are then used for generating the compressed data which should be stored in the hardware devices. It is observed that these proposed algorithms could achieve 32.46% compression ratio averagely.

Many of the standard compression methods are based, at their lowest level, in coding digital words of variable length or VLC, Huffman type, designed in 1952 or in any of its versions, as canonical. This article presents an VLC with greater efficiency than Huffman encoding, named as “Leveling”, which uses two variants, “Leveled Reordering” for low redundancy, e.g. for text and, “Segmented Leveling” for middle and high redundancy, for image processing. Leveling, developed by Javier Joglar in 1995, uses the concepts of “meaning” and “ordering” of the VLC codes generated, to get optimum performance in terms of “compression ratio”, higher than any other non-adaptive VLC.

Swing door trending (SDT) algorithm is a lossy compression algorithm that be applied on the real-time database and proposed by OSI software company of American; SDT is widely used to compress process data generated by process industry. Using straight line for a data section to linear fitting in traditional SDT algorithm. However, the data generated in the process of industrial production are slightly fluctuating with time. So, the use of linear fitting will lead to a large decompression error. In order to overcome the large decompression error generated by the traditional SDT, we proposed Controllable Curve Fitting Based Swing Door Trending (CCFSDT). The CCFSDT algorithm uses curve line for a data section to fitting, the data restored are closer to the true value. And in order to reduce the cost of curve fitting, it can be appropriate to reduce the total number of points of curve fitting. We filter noise point before fitting to void the impact on the reduction data and achieve better compression effect. The experimental results on simulated data and actual plant data show that: under the same conditions, the CCFSDT can well reduce the errors of decompression and achieve satisfactory performance.

In this paper, it is considered the image compression scheme, in which a part of coded data extracted in the coding process is hidden into the other parts of coded data of own image, especially into the block address data of the best matching block within restricted blocks. The proposed scheme is able to be used in a fractal image coding in which the best matching domain block is searched, in a vector quantization in image coding in which the best matching vector is searched, and in motion compensation of moving picture in which the best matching motion vector is searched. We study each image coding method and consider the features of each coding method using the proposed scheme.

DNA sequences comprise of sequentially linked nucleotides, A, C, G and T. As a result of the genome projects, a significant amount of DNA sequences of various species are deposited in various databases. Human DNA contains about 3 billion base pairs. The number of genes within the DNA is 20,000 to 25,000. For storing DNA data of a single person, we require approximately 10 CD – ROMs. This amounts to huge data storage costs, subsequently making the use of these data such as analysis and retrieval quite challenging. DNA sequence analysis is useful in diverse areas such as forensics, medical research, pharmacy, agriculture etc. It is very necessary to address the storage issue of these exponentially growing data. In this paper we have implemented 4 different algorithms for DNA data compression: LZW (Lampel-ziv-Welch) algorithm, run length encoding algorithm, Arithmetic coding and Substitution method. The compression results on these algorithms are presented and compared on DNA sequence data of 10 different species.

Most time series big data is with noise and uncertain. To abstract the key information effectively and quickly, the estimation is one of the feasible methods for the uncertain big data. The Kalman filter with adaptive method by part of samples can give the high dimensional characteristics, reduce the computing cost and data uncertainty, but encounter the irregular estimation. The number of sample and the performance of the abstracted information have the tradeoff, which means we can use the suitable number of sample to abstract the key information of the series data. This paper discusses how to find the suitable sampling points for the time series data and the simulations show that the key dynamic information of time series big data can be guaranteed with the compression amount number of sample data.