Clean Ones Research Articles

Cryptocurrency and Money Laundering

Cryptocurrency, a form of virtual currency, is increasingly pervasive in the modern society. People can use Bitcoin, Ether or Dogecoin to buy a range of products and services with ease and convenience. Yet, beneath the popularity of cryptocurrency is an emerging socio-legal concern: money laundering. Using the information collected from web sources, this brief paper taps into how cryptocurrency money laundering works and what measures can be undertaken to curtail digital crime. In view of the extant information, the paper found that a mixer is required as a ‘middle man’ to convert identifiable crypto-tokens into unidentifiable clean ones thereby delivering them to new wallet(s). In this manner, the origins of these tokens can be obscured, and the new tokens can be accessed and used legally. Many platforms and channels can serve as the mixer to proceed with illegal activities, including casinos, dark web marketplaces, and p2p networks. In considering the ways to tackle such digital crime in jurisdictions with limited oversight, the paper proposes to adopt measures in the Anti-Money Laundering initiatives already implemented in Hong Kong, the United States, and Singapore – especially, the need to strengthen recordkeeping to enhance the traceability and trackability of cryptocurrencies. Meanwhile, laws and rules associated with digital crime shall also be reshaped for risk mitigation.

MetaSeg: Content-Aware Meta-Net for Omni-Supervised Semantic Segmentation.

Noisy labels, inevitably existing in pseudo-segmentation labels generated from weak object-level annotations, severely hamper model optimization for semantic segmentation. Previous works often rely on massive handcrafted losses and carefully tuned hyperparameters to resist noise, suffering poor generalization capability and high model complexity. Inspired by recent advances in meta-learning, we argue that rather than struggling to tolerate noise hidden behind clean labels passively, a more feasible solution would be to find out the noisy regions actively, so as to simply ignore them during model optimization. With this in mind, this work presents a novel meta-learning-based semantic segmentation method, MetaSeg, that comprises a primary content-aware meta-net (CAM-Net) to serve as a noise indicator for an arbitrary segmentation model counterpart. Specifically, CAM-Net learns to generate pixel-wise weights to suppress noisy regions with incorrect pseudo-labels while highlighting clean ones by exploiting hybrid strengthened features from image content, providing straightforward and reliable guidance for optimizing the segmentation model. Moreover, to break the barrier of time-consuming training when applying meta-learning to common large segmentation models, we further present a new decoupled training strategy that optimizes different model layers in a divide-and-conquer manner. Extensive experiments on object, medical, remote sensing, and human segmentation show that our method achieves superior performance, approaching that of fully supervised settings, which paves a new promising way for omni-supervised semantic segmentation.

Heat-altered scent marks of males of a fossorial reptile still allow recognition by females but lose information on male quality

The efficacy of sexual signals in communication is often maximized under specific environmental conditions. Anthropic alterations of these conditions might, thus, negatively affect communication during reproductive behavior. In fossorial animals, inhabiting visually restricted environments, chemical senses are very important. We examined whether climatic episodes of heat-waves with unusual high temperatures may affect the information provided to females by the sexual chemical signals of males of a fossorial reptile, the amphisbaenian Blanus cinereus. The results showed that experimentally heat-altered substrate scent marks of males can still provide information to females about the presence of a male. Females spent more time on males’ scent marks, irrespective of the temperature treatment, than on control clean ones. However, heat-altered scent marks did not seem to convey information about the health state (immune response) of the producer. Females spent more time on unaltered scent marks of healthier males (probably indicating mating preferences for these males), while female preferences for some heat-altered scent marks were not related to size or immune response of the same individual males. Chemical analyses indicated that the overall chemical profile of precloacal secretions (used for scent marking) did not change with increased temperatures. However, the relationship between proportions of some compounds in secretions and males’ immune response found in unaltered secretions was lost in heat-altered ones. We conclude that unusual increased environmental temperatures may decrease the efficacy of underground sexual chemical signals in this amphisbaenian (i.e., a loss of information on male quality), and consequently, may negatively affect sexual selection and reproduction.

Regroup Median Loss for Combating Label Noise

The deep model training procedure requires large-scale datasets of annotated data. Due to the difficulty of annotating a large number of samples, label noise caused by incorrect annotations is inevitable, resulting in low model performance and poor model generalization. To combat label noise, current methods usually select clean samples based on the small-loss criterion and use these samples for training. Due to some noisy samples similar to clean ones, these small-loss criterion-based methods are still affected by label noise. To address this issue, in this work, we propose Regroup Median Loss (RML) to reduce the probability of selecting noisy samples and correct losses of noisy samples. RML randomly selects samples with the same label as the training samples based on a new loss processing method. Then, we combine the stable mean loss and the robust median loss through a proposed regrouping strategy to obtain robust loss estimation for noisy samples. To further improve the model performance against label noise, we propose a new sample selection strategy and build a semi-supervised method based on RML. Compared to state-of-the-art methods, for both the traditionally trained and semi-supervised models, RML achieves a significant improvement on synthetic and complex real-world datasets. The source is at https://github.com/Feng-peng-Li/Regroup-Loss-Median-to-Combat-Label-Noise.

A GAN Guided Parallel CNN and Transformer Network for EEG Denoising.

Electroencephalography (EEG) signals are often contaminated with various physiological artifacts, seriously affecting the quality of subsequent analysis. Therefore, removing artifacts is an essential step in practice. As of now, deep learning-based EEG denoising methods have exhibited unique advantages over traditional methods. However, they still suffer from the following limitations. The existing structure designs have not fully taken into account the temporal characteristics of artifacts. Meanwhile, the existing training strategies usually ignore the holistic consistency between denoised EEG signals and authentic clean ones. To address these issues, we propose a GAN guided parallel CNN and transformer network, named GCTNet. The generator contains parallel CNN blocks and transformer blocks to respectively capture local and global temporal dependencies. Then, a discriminator is employed to detect and correct the holistic inconsistencies between clean and denoised EEG signals. We evaluate the proposed network on both semi-simulated and real data. Extensive experimental results demonstrate that GCTNet significantly outperforms state-of-the-art networks in various artifact removal tasks, as evidenced by its superior objective evaluation metrics. For example, in the task of removing electromyography artifacts, GCTNet achieves 11.15% reduction in RRMSE and 9.81% improvement in SNR over other methods, highlighting the potential of the proposed method as a promising solution for EEG signals in practical applications.

Progressive dense feature fusion network for single image deraining

Deep Learning (DL) has achieved significant progress in single image deraining methods. Most of current DL methods, however, are still weak in image detail recovery and feature inherent correlation learning. In this work, we explore the detail recovery mechanisms in both network architecture and loss function for image deraining. A U-net like architecture named progressive dense feature fusion network (PDFFN) is proposed to encode rain images and decode them as clean ones. Specifically, a residual dense connection unit (ReDCU) is designed to handle rain streaks of various blurring degrees and resolutions by enriching features. Moreover, a progressive feature fusion module (PFFM), which fuses the features in different stages of U-net progressively, is devised to not only capture the inherent correlations of features across encoder and decoder but also intensify the fine-gained details. To better evaluate the perceptual similarity between the ground-truth and derained image, we propose detail perceptual loss by focusing on low-level unactivated features. Apart from the global rain removal strategy, this paper applies contextual loss on target regions to conduct joint deraining and detection. Comprehensive experiments substantiate the superiority of the proposed method, especially its detail recovery capability, as compared with state-of-the-art methods both qualitatively and quantitatively.

The influence of household characteristics on the purchase of clean cars. The case of Spain

This paper determines the socioeconomic characteristics of Spanish households that use clean cars, with the aim of identifying how these characteristics influence the market penetration of these vehicles. This study is one of the few in the literature that analyses the characteristics of car users based on a broad sample of data, that is representative of 11.5 million households. The results allow us to identify socioeconomic challenges and opportunities regarding the goal of replacing combustion vehicles with clean vehicles. We estimate a logit model finding that high-income households are 29.3% more likely to use a clean vehicle than medium–low-income households. On the contrary, the gender and age of the household’s main breadwinner present opportunities: if he is a male, it is more likely to use clean vehicles than if she is a woman, and households whose main breadwinner is over 55 or under 30 are more likely to use it. These results provide guidelines for a better design of public policies aimed at decarbonizing the transport sector through the replacement of conventional vehicles by clean ones. To do so it is important to consider the features of the social reality in which that objective is to be achieved.

Does inward remittance influence the choice of households’ cooking fuel?

ABSTRACT Among the Sustainable Development Goals is to ensure the consumption of clean and affordable energy by all by the close of 2030. Over the last decade, calls for energy transition from unclean energy to clean ones have intensified. Consequently, the price of clean energy has been subsidised in Ghana but this is not enough as energy transition still entails considerable costs for households. The receipt of inward remittance improves households’ income which can aid their transition but literature in this regard is scanty. Applying the Special Regression Method to household data from the Ghana Living Standards Survey, we examine how the receipt of income remittance influences the choice of households’ cooking fuels. We found an increase in income remittance increases the odds of households adopting clean cooking fuels. We recommend that the government of Ghana undertake policies that improve household income to encourage the adoption of clean fuels and reduce environmental pollution.

The Green Effect: Exploring the Impact of Innovation and Foreign Direct Investment on Environmental Quality in Malaysia

Do innovation and foreign direct investment (FDI) matter in greening the eco-system, and why not? They can lead to new technologies and knowledge, better efficient renewable energy, and reduced energy consumption; in turn this presumably reduces carbon emissions. In this study, we capture the short-and long-term impact of innovation and FDI, among other macro-variables, on the environmental quality in Malaysia using time series data over the period 1990-2020. Our empirical models find that both innovation and FDI have significant and positive impact on environmental quality in Malaysia, although it is only significant in the short-run for the first one, while significant in the long-run for the latter ones. The disappointing findings from our study is that both economic growth and urbanization show negative impact on environmental quality as CO2 tends to increase with higher levels of GDP and urbanization. An important policy recommendation that we can draw from this paper is that Malaysian policy-makers should, and foremost, focus on designing growth and urbanization policies, regulations and procedures that are environmentally induced to achieve decoupling goal in its development path. We believe that attracting more focused FDI that supports clean and green economy, better innovations and policies to replace traditional energy with clean and renewable ones, could achieve decoupling goal that leads to a reduction in CO2 emissions and improve environmental quality while keep growing in the world economy, including Malaysia.

A Novel Adversarial Learning Framework for Passive Bistatic Radar Signal Enhancement

Passive Bistatic Radar (PBR) has significant civilian and military applications due to its ability to detect low-altitude targets. However, the uncontrollable characteristics of the transmitter often lead to subpar target detection performance, primarily due to a low signal-to-noise ratio (SNR). Coherent accumulation typically has limited ability to improve SNR in the presence of strong noise and clutter. In this paper, we propose an adversarial learning-based radar signal enhancement method, called radar signal enhancement generative adversarial network (RSEGAN), to overcome this challenge. On one hand, an encoder-decoder structure is designed to map noisy signals to clean ones without intervention in the adversarial training stage. On the other hand, a hybrid loss constrained by L1 regularization, L2 regularization, and gradient penalty is proposed to ensure effective training of RSEGAN. Experimental results demonstrate that RSEGAN can reliably remove noise from target information, providing an SNR gain higher than 5 dB for the basic coherent integration method even under low SNR conditions.

An experimental investigation of hydrogen-enriched and nanoparticle blended waste cooking biodiesel on diesel engine

The hazardous effect of the pollution of fossil fuels has brought the necessity of shifting conventional energy sources to renewable and clean ones. In this study, the effect of hydrogen addition and CeO2 nanoparticle addition in waste cooking palm biodiesel on a CRDI engine is evaluated. The dosage of the nanoparticle is fixed at 75 ppm and a hydrogen flow rate of 10 L/min is selected for the engine operations. The crystalline structure of the nanoparticles is determined by XRD analysis. Results showed that on the addition of both H2 and CeO2 in a B20 biodiesel blend (80% diesel and 20% biodiesel) the performance, emission, and combustion parameters of the diesel engine improved compared to neat diesel. The brake thermal efficiency was improved by 3.53% and brake fuel consumption was reduced by 16.12% in comparison to diesel at 90% loading condition. The addition of both nanoparticles and hydrogen in the biodiesel blend lowered the emissions of CO by 30%, and HC and smoke by 50% and 42% respectively. However, NOx increased by 11% as compared to diesel. A 6% higher HRR values and 8% higher in-cylinder pressure were obtained while using hydrogen and CeO2 nanoparticle blended biodiesel. This blend also shows the lowest ignition delay period at full load condition which results in more engine power and efficiency. This experimental study has helped pave the way for the use of hydrogen-enriched and nanoparticle-blended biodiesel in place of fossil fuel for the applications of diesel engines.

A Segmentation-Denoising Network for Artifact Removal From Single-Channel EEG

As an important neurorecording technique, electroencephalography (EEG) is often contaminated by various artifacts, which obstructs subsequent analysis. In recent years, deep learning-based methods have been proven to be promising for artifact removal. However, most denoising methods focus on recovering clean EEG from raw signals contaminated by the noise over the entire recording period, ignoring that the practical EEG recordings may contain clean segments in addition to noise segments. Therefore, the general model may cause distortion when dealing with clean segments. In this paper, we propose a simple, yet effective segmentation-denoising network (SDNet) for artifact removal. The proposed method is capable of differentiating noisy EEG segments from clean ones via semantic segmentation, avoiding the distortion caused by processing clean segments. We conduct a performance comparison on semi-simulated and real EEG data. The experimental results demonstrate that SDNet outperforms the state-of-art approaches. This work provides a novel way to reconstruct artifact-attenuated EEG signals, and may further benefit the EEG-based diagnosis and treatment.

Co-mining: Mining informative samples with noisy labels

Due to the memorization effect of Deep Neural Networks, training with noisy labels typically results in performance degradation. Existing methods have devoted numerous efforts to training a robust model. However, rare literature pays attention to deep metric learning in the presence of label noise due to the challenge of discriminating between noisy examples and informative (hard) ones. To this end, we propose a simple yet effective approach Co-mining to build a robust and discriminative feature space. Specifically, more confident yet fewer instances are selected to guarantee the label quality and make up a relatively reliable gallery for exploring under-exploited patterns. Considering that some unexpected noisy instances are inevitably registered as clean ones as training progresses, we make a trade-off between cleanliness and informativeness. A triplet loss, pulling positive pairs closer and pushing negative pairs apart, is imposed on semi-hard samples rather than the hardest counterparts. Lastly, our method employs a threshold to distinguish out-of-distribution noisy images and discards them directly. We also use a re-ranking strategy for in-distribution noisy data to correct corrupted labels based on their consistently high predicted probabilities. Comprehensive experiments on various synthetic and real-world benchmarks demonstrate the superiority of our Co-mining. Code is available at https://github.com/NUST-Machine-Intelligence-Laboratory/Co-mining.

Do FDI Inflows into African Countries Impact Their CO2 Emission Levels?

The emitted levels of CO2 continue to be a striking topic. These emissions have been growing over the years, thus, making them a predicament to be reckoned with. Eradicating such a predicament has not been easy because finding an optimal determinant has not been achieved by scholars; however, foreign direct investment inflows are known to play a role in such varying instances. Therefore, to analyze the impact that such inflows have on CO2 emissions, this study employs data from 41 African countries from 2005 to 2019 and aims to assess how foreign direct investment and other variables influence CO2 emitted levels. Moreover, this study tests the validity of the pollution haven and halo hypotheses on the employed African countries as its two main objectives. After applying the pooled least squares, fixed and random effects models, and the generalized method of moments, the findings revealed that per the adopted African countries, the pollution haven and halo hypotheses do not hold; however, foreign direct investment inflows contribute to the rising and falling levels of CO2 emissions. In addition, the financial structure and per capita GDP increase the African countries’ CO2 emitted levels, while trade openness causes a reduction. Based on the aforementioned findings, this study recommends that the government, policy-makers, industries, and interested personnel of this study’s employed countries should: apply and execute policies, laws, and regulations that will deter or punish polluting foreign investment and encourage clean ones; since green finance is making waves but is not well established in most African countries, green financing systems should be initiated and implemented; establish preferential trading policies that will highlight an addition of value via clean technology; and practice carbon capture, usage, and storage.

Data Expansion Approach with Attention Mechanism for Learning with Noisy Labels

In recent years, the development of deep learning has contributed to various areas of machine learning. However, deep learning requires a huge amount of data to train the model, and data collection techniques such as web crawling can easily generate incorrect labels. If a training dataset has noisy labels, the generalization performance of deep learning significantly decreases. Some recent works have successfully divided the dataset into samples with clean labels and ones with noisy labels. In light of these studies, we propose a novel data expansion framework to robustly train the models on noisy labels with the attention mechanisms. First, our method trains a deep learning model with the sample selection approach and saves the samples selected as clean at the end of training. The original noisy dataset is then extended with the selected samples and the model is trained on the dataset again. To prevent over-fitting and allow the model to learn different patterns of the selected samples, we leverage the attention mechanism of deep learning to modify the representation of the selected samples. We evaluated our method with synthetic noisy labels on CIFAR-10 and CUB-200-2011 and real-world dataset Clothing1M. Our method obtained comparable results to baseline CNNs and state-of-the-art methods.

Deep Learning Algorithm as an Alternative of Automatic Classification of Dirty Eggs

Egg, which is one of the daily needs, contains important nutrition like protein, vitamins, mineral and 75 Kcal. Thus, one of the external qualities of eggs becomes one of the considerations in determining egg quality. The poor condition of the egg shell can be unhealthy and harmful. The dirty eggs and the clean ones are important elements in an automatic separation system since they relate to the consumer’s economic and health values. An element such as dirt at the eggshell leads the raise of Salmonella bacteria and will be worse if the bacteria get into the eggshell. An automatic method of egg selection is crucial to determining the egg's quality before it reaches the consumer. Selecting and classifying egg manually is inefficient for the seller, due to it takes more time and complicated processes. The development of an automatic method is required to categorize and simplify the manual system. Deep Learning Algorithm become one of remarkable solution in developing the method of classification and qualified egg. The method of deep learning algorithms is a computational process that is simpler than other computing methods. The process of egg selection is conducted based on the dirtiness of the eggs. The result of this system is able to classify accurately and efficiently to the select whether the egg has good or bad quality, in order to simplify the breeders’ job.

ARU-DGAN: A dual generative adversarial network based on attention residual U-Net for magneto-acousto-electrical image denoising.

Magneto-Acousto-Electrical Tomography (MAET) is a multi-physics coupling imaging modality that integrates the high resolution of ultrasound imaging with the high contrast of electrical impedance imaging. However, the quality of images obtained through this imaging technique can be easily compromised by environmental or experimental noise, thereby affecting the overall quality of the imaging results. Existing methods for magneto-acousto-electrical image denoising lack the capability to model local and global features of magneto-acousto-electrical images and are unable to extract the most relevant multi-scale contextual information to model the joint distribution of clean images and noise images. To address this issue, we propose a Dual Generative Adversarial Network based on Attention Residual U-Net (ARU-DGAN) for magneto-acousto-electrical image denoising. Specifically, our model approximates the joint distribution of magneto-acousto-electrical clean and noisy images from two perspectives: noise removal and noise generation. First, it transforms noisy images into clean ones through a denoiser; second, it converts clean images into noisy ones via a generator. Simultaneously, we design an Attention Residual U-Net (ARU) to serve as the backbone of the denoiser and generator in the Dual Generative Adversarial Network (DGAN). The ARU network adopts a residual mechanism and introduces a linear Self-Attention based on Cross-Normalization (CNorm-SA), which is proposed in this paper. This design allows the model to effectively extract the most relevant multi-scale contextual information while maintaining high resolution, thereby better modeling the local and global features of magneto-acousto-electrical images. Finally, extensive experiments on a real-world magneto-acousto-electrical image dataset constructed in this paper demonstrate significant improvements in preserving image details achieved by ARU-DGAN. Furthermore, compared to the state-of-the-art competitive methods, it exhibits a 0.3 dB increase in PSNR and an improvement of 0.47% in SSIM.

Learning a Task-Specific Descriptor for Robust Matching of 3D Point Clouds

Existing learning-based point feature descriptors are usually task-agnostic, which pursue describing the individual 3D point clouds as accurate as possible. However, the matching task aims at describing the corresponding points consistently across different 3D point clouds. Therefore these too accurate features may play a counterproductive role due to the inconsistent point feature representations of correspondences caused by the unpredictable noise, partiality, deformation, \etc, in the local geometry. In this paper, we propose to learn a robust task-specific feature descriptor to consistently describe the correct point correspondence under interference. Born with an Encoder and a Dynamic Fusion module, our method EDFNet develops from two aspects. First, we augment the matchability of correspondences by utilizing their repetitive local structure. To this end, a special encoder is designed to exploit two input point clouds jointly for each point descriptor. It not only captures the local geometry of each point in the current point cloud by convolution, but also exploits the repetitive structure from paired point cloud by Transformer. Second, we propose a dynamical fusion module to jointly use different scale features. There is an inevitable struggle between robustness and discriminativeness of the single scale feature. Specifically, the small scale feature is robust since little interference exists in this small receptive field. But it is not sufficiently discriminative as there are many repetitive local structures within a point cloud. Thus the resultant descriptors will lead to many incorrect matches. In contrast, the large scale feature is more discriminative by integrating more neighborhood information. ...

Monitoring the security of audio biomedical signals communications in wearable IoT healthcare

The COVID-19 pandemic has imposed new challenges on the healthcare industry as hospital staff are exposed to a massive coronavirus load when registering new patients, taking temperatures, and providing care. The Ebola epidemic of 2014 is another example of a pandemic which a hospital in New York decided to use an audio-based communication system to protect nurses. This idea quickly turned into an Internet of Things (IoT) healthcare solution to help to communicate with patients remotely. However, it has grabbed the attention of criminals who use this medium as a cover for secret communication. The merging of signal processing and machine-learning techniques has led to the development of steganalyzers with very higher efficiencies, but since the statistical properties of normal audio files differ from those of purely speech audio files, the current steganalysis practices are not efficient enough for this type of content. This research considers the Percent of Equal Adjacent Samples (PEAS) feature for speech steganalysis. This feature efficiently discriminates the least significant bit stego speech samples from clean ones with a single analysis dimension. A sensitivity of 99.82% was achieved for the steganalysis of 50% embedded stego instances using a classifier based on the Gaussian membership function.

A Real-Time Automated Defect Classification System Based on Two-Step Convolutional Neural Network

The increasing amount of defect images in semiconductor manufacturing process imposes the demand of classifying these images in real-time. In this paper we adopt a two-step deep learning based Convolutional Neural Network (CNN) to detect and classify the defect images from clean ones. The proposed method integrates the detection and classification process into one forward step, which takes image-level feature as well as empirical defect classification rules into account. In practice, the Real-Time Automated Defect Classification (RT-ADC) system was deployed in inline production process, which brings high efficiency; better wafer analysis and root cause determination for yield enhancement.