Local Machine Research Articles

Analysis and Detection of Fraud in Credit Card Using SILOF

Nowadays e-commerce and online transaction is growing rapidly. For online and offline transaction most of the customer uses credit card. Credit card used globally for online transaction, buy goods, product, and payment. The rising use of credit card can increase the chances of fraud in credit card. Credit card system is at risk now. The effect of this fraudulent transaction is on the bank and institute causing a financial loss to them. For the detection of distinguish frauds, several machine learning models are utilized for better prediction. The major objective of this article is to identify the fraudulent transaction and outlier in credit card transaction. The dataset of credit card is unbalanced. There are various techniques by which fraudulent transaction can be detected and we have used these techniques such as isolation forest method, local outlier factor and support vector machine to determine fraud in credit card. We have used different matrices for enhancing the performance and accuracy. At last comparison analysis is done by using isolation forest, support vector machine, and local outlier which give the better result.

Reframing Demand Forecasting: A Two-Fold Approach for Lumpy and Intermittent Demand

Demand forecasting is a crucial component of demand management. While shortening the forecasting horizon allows for more recent data and less uncertainty, this frequently means lower data aggregation levels and a more significant data sparsity. Furthermore, sparse demand data usually result in lumpy or intermittent demand patterns with irregular demand intervals. The usual statistical and machine learning models fail to provide good forecasts in such scenarios. Our research confirms that competitive demand forecasts can be obtained through two models: predicting the demand occurrence and estimating the demand size. We analyze the usage of local and global machine learning models for both cases and compare the results against baseline methods. Finally, we propose a novel evaluation criterion for the performance of lumpy and intermittent demand forecasting models. Our research shows that global classification models are the best choice when predicting demand event occurrence. We achieved the best results using the simple exponential smoothing forecast to predict demand sizes. We tested our approach on real-world data made up of 516 time series corresponding to the daily demand, over three years, of a European original automotive equipment manufacturer.

Why Allow Local Elections? Mobilization, Manipulation, and the Abolition of Russian Mayoral Elections

Abstract Local executives in electoral authoritarian regimes can perform important regime-sustaining functions, including by delivering votes to the ruling party at election-time. Furthermore, when local executives are themselves elected, regimes can benefit from improved legitimacy and efficiency in local government. Yet elected local executives can create principal-agent problems and increase the risk that opposition groups gain office. How do authoritarian governments manage this tension? Prior research on Russia shows that elections are used to co-opt strong local mayors, while weak mayors are replaced with appointed managers. This paper argues that strong mayors are more likely to see elections canceled if their local machine is not delivering manufactured electoral support to the national party, while weak mayors are unlikely to be targeted. This hypothesis is supported using data from 207 Russian cities, including election-forensic estimates of election manipulation. The findings improve our understanding of cooptation of local leaders in electoral authoritarian regimes.

A Differential Approach for Data and Classification Service-Based Privacy-Preserving Machine Learning Model in Cloud Environment

The massive upsurge in computational and storage has driven the local data and machine learning applications to the cloud environment. The owners may not fully trust the cloud environment as it is managed by third parties. However, maintaining privacy while sharing data and the classifier with several stakeholders is a critical challenge. This paper proposes a novel model based on differential privacy and machine learning approaches that enable multiple owners to share their data for utilization and the classifier to render classification services for users in the cloud environment. To process owners data and classifier, the model specifies a communication protocol among various untrustworthy parties. The proposed model also provides a robust mechanism to preserve the privacy of data and the classifier. The experiments are conducted for a Naive Bayes classifier over numerous datasets to compute the proposed model efficiency. The achieved results demonstrate that the proposed model has high accuracy, precision, recall, and F1-score up to 94%, 95%, 94%, and 94%, and improvement up to 16.95%, 20.16%, 16.95%, and 23.33%, respectively, compared with state-of-the-art works.

Building Trusted Federated Learning on Blockchain

Federated learning enables multiple users to collaboratively train a global model using the users’ private data on users’ local machines. This way, users are not required to share their training data with other parties, maintaining user privacy; however, the vanilla federated learning proposal is mainly assumed to be run in a trusted environment, while the actual implementation of federated learning is expected to be performed in untrusted domains. This paper aims to use blockchain as a trusted federated learning platform to realize the missing “running on untrusted domain” requirement. First, we investigate vanilla federate learning issues such as client’s low motivation, client dropouts, model poisoning, model stealing, and unauthorized access. From those issues, we design building block solutions such as incentive mechanism, reputation system, peer-reviewed model, commitment hash, and model encryption. We then construct the full-fledged blockchain-based federated learning protocol, including client registration, training, aggregation, and reward distribution. Our evaluations show that the proposed solutions made federated learning more reliable. Moreover, the proposed system can motivate participants to be honest and perform best-effort training to obtain higher rewards while punishing malicious behaviors. Hence, running federated learning in an untrusted environment becomes possible.

Performance Evaluation of Local Design Machine (LDSM Type) on Productivity and Growth of Wheat

A field experiment was conducted to assess the influence of agricultural machines (AM) LDSM and Famarol, and planting distances (CD) on wheat/ Rashed cultivar. The experiment included two types of AM (LDSM and Famarol) and three CD levels (CD of 10, CD of 12 and CD of 15cm) in three application. The split -split plot arrangement in RCBD with three replications was used, the results indicated that the AM (Famarol type) was major impact than AM (LDSM type) in all studied conditions, The CD of 10 cm, was significantly superior to the other two levels of CD of 12 cm, and CD of 15 cm for most growth characters. PVI, RL, HI and GY, except FC which gave best result with CD of 15cm.

Privacy-Preserving Federated Edge Learning: Modeling and Optimization

In this letter, we consider the personalized differential privacy (DP) based federated edge learning system. Each edge device adds DP noise to its local machine learning (ML) model updates to prevent the private information contained in the model updates to be obtained by the edge server. However, the noise perturbation can degrade the ML model performance. We aim to optimize the tradeoff between the ML model performance measured by the global loss and the privacy preservation. The closed-form global loss and privacy leakage are first derived. The loss and leakage are then jointly minimized by optimizing the DP noise scales and the local update numbers of the edge devices. Numerical results show that a better loss-leakage trade-off is reached compared to the conventional methods.

Enhancing the Mechanical Characteristics of the Traditional Concrete with the Steel Scrap

The mechanical performance of concrete with varying proportions of steel scraps as a composite additive is investigated in this work. A M50 grade concrete admixture with a 0.35 water-to-binder ratio was prepared for this study. The appropriate quantity of superplasticizer was included as the mineral admixture. Steel scraps are obtained as waste from local machining workshops and then included in various proportions, including 0.25, 0.5, 1.0, and 2.0 percentages. The cubical mortar specimens were being employed to determine the compressive characteristics of mortar structure with and without steel scrap, whilst the cylinder-formed specimens and beam shaped samples were being utilized to determine the indirect tensile strength and flexural strength of concrete mixture. All experiments with different percentages of steel scrap were conducted on the 3 replicates, and the mean value is provided in this paper. After 28 days, the cement without steel scrap seemed to have a mean compression, flexural, and tensile strengths of 46.3 MPa, 5.52 MPa, and 4.23 MPa, respectively, which were improved to 51.7 MPa, 6.16 MPa, and 4.58 MPa with the inclusion of 1.0% steel scrap. The proposed investigation will contribute to reducing cement use, hence reducing cement industry's adverse ecological impacts.

A Wave Beam Interlocking Model and Simulation Based on Position Information Interaction

Aiming at the problem of wave beam interlock in directional antenna communication of aviation swarm, a mathematical model of wave beam interlock based on position information interaction is established. The core idea of this model is to transform the position information of the local machine and other machines into the wave beam direction of directional antenna through a series of coordinate transformations, so as to realize wave beam interlocking. This paper analyzes the source of the precision error of wave beam interlock in mathematical model, emphatically demonstrates the precision error of attitude measurement, and gives the error boundary conditions to maintain normal communication. Finally, the computer simulation of the mathematical model and error sources proves the validity of the mathematical model and the correctness of the error analysis. The model is simple, reliable and practical.

Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication

In this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound.

Learning the Quantum Centroid Force Correction in Molecular Systems: A Localized Approach.

Molecular mechanics (MM) is a powerful tool to study the properties of molecular systems in the fields of biology and materials science. With the development of ab initio force field and the application of ab initio potential energy surface, the nuclear quantum effect (NQE) is becoming increasingly important for the robustness of the simulation. However, the state-of-the-art path-integral molecular dynamics simulation, which incorporates NQE in MM, is still too expensive to conduct for most biological and material systems. In this work, we analyze the locality of NQE, using both analytical and numerical approaches, and conclude that NQE is an extremely localized phenomenon in nonreactive molecular systems. Therefore, we can use localized machine learning (ML) models to predict quantum force corrections both accurately and efficiently. Using liquid water as example, we show that the ML facilitated centroid MD can reproduce the NQEs in both the thermodynamical and the dynamical properties, with a minimal increase in computational time compared to classical molecular dynamics. This simple approach thus largely decreases the computational cost of quantum simulations, making it really accessible to the studies of large-scale molecular systems.

First liquid chromatography-high resolution mass spectrometry method for the determination of cocaine on banknote dust.

Prevalence measures of sociological interest concerning cocaine presence on banknotes are fraught with (i) the extreme variability of its concentration (seven orders of magnitude); (ii) the high number of banknotes needed for the statistical significance. Banknote dust from counting machines from a large and representative number of banknotes in circulation in a specific area represents the most eligible sample to ascertain cocaine circulation. No chromatographic method is available in this respect. This study aims at developing the first analytical methodology for the determination of cocaine in banknote dust samples. This novel and straightforward approach consists of a simple methanol extraction followed by analytical determinations via ultra-high performance liquid chromatography coupled to Orbitrap high-resolution mass spectrometry. Satisfactory analytical performance was obtained with a coefficient of determination of 0.996; maximum within-run and between-run precisions were, respectively, 1.85% and 5.20%. Limits of detection and quantification were, respectively, 3 and 9ng/mL with an overall process efficiency of 93.2%. The method developed was successfully applied to 9 banknote dust samples from local banknote counter machines. The found concentrations ranged from 2.18E + 02 to 2.31E + 03μg of cocaine per gram of banknote dust and varied only one order of magnitude, much less than cocaine concentration on banknotes. To have an idea of cocaine circulation in a geographical area, the sampling of banknote dust, compared to banknotes, consists of tremendous advantages in terms of statistical significance, higher cocaine concentrations, and lower variability: this is crucial from the sociological point of view.

Multipath Transmission Control Protocol for Live Virtual Machine Migration in the Cloud Environment

For mobile cloud computing (MCC), a local virtual machine- (VM-) based cloudlet is proposed to improve the performance of real-time resource-intensive mobile applications. When a mobile device (MD) discovers a cloudlet nearby, it takes some time to build up a virtual machine (VM) inside the cloudlet before data offloading from the MD to the VM can begin. Live virtual machine migration refers to the process of transferring a running Virtual Machine (VM) from one host to another without interrupting its state. Theoretically, live migration process must not render the instance being migrated unavailable during its execution. However, in practice, there is always a service downtime associated with the process. This paper focuses on addressing the need to reduce the service downtime in case of live VM migration in cloud and providing a solution by implementing and optimizing the multipath transmission control protocol (MPTCP) ability within an Infrastructure as a service (IaaS) cloud to increase the efficiency of live migration. We have also introduced an algorithm, the α-best fit algorithm, to optimize the usage of bandwidth and to effectively streamline the MPTCP performance.

An Interethnic Paradox: Chicago's Irish and Everyone Else

An Interethnic Paradox: Chicago's Irish and Everyone Else

A localized DNA finite-state machine with temporal resolution.

The identity and timing of environmental stimulus play a pivotal role in living organisms in programming their signaling networks and developing specific phenotypes. The ability to unveil history-dependent signals will advance our understanding of temporally regulated biological processes. Here, we have developed a two-input, five-state DNA finite-state machine (FSM) to sense and record the temporally ordered inputs. The spatial organization of the processing units on DNA origami enables facile modulation of the energy landscape of DNA strand displacement reactions, allowing precise control of the reactions along predefined paths for different input orders. The use of spatial constraints brings about a simple, modular design for the FSM with a minimum set of orthogonal components and confers minimized leaky reactions and fast kinetics. The FSM demonstrates the capability of sensing the temporal orders of two microRNAs, highlighting its potential for temporally resolved biosensing and smart therapeutics.

Improving blade accuracy via local electrochemical machining with partial insulated cathodes

Electrochemical machining (ECM) has memory effect characteristics, which transfers the initial allowance unevenness to the final product, especially in conditions of small allowances such as in blades and blisks. This affects the machining accuracy and can lead to local overcutting. Thus, a two-step blade ECM method is proposed to solve this problem. The step 1 is local precise modifications using partial insulated electrodes, and. the step 2 is contour machining using conventional standard cathodes. Simulation investigations are performed to obtain different allowance distributions under various insulation widths. The relationship between the allowance distribution and insulation width dictates the most suitable insulation width in the step 1. To verify the effects of this method in blade machining with small allowances, experiments between the one-step and two-step blade ECM methods are compared. The experimental results show that the blade is overcut by 0.43 mm in the traditional one-step blade ECM method. However, for the two-step method, the allowance difference is reduced from 3.97 mm to 0.22 mm in the step 1, and a blade with the profile machining accuracy of −0.02–0.04 mm is machined after the step 2. Thus, the experimental results verify that the two-step blade ECM method is effective for blade machining with small allowances.

Improving TVB-N prediction in pork using portable spectroscopy with just-in-time learning model updating method

Near infrared spectroscopy (NIR) technology is an effective method for nondestructive prediction of total volatile basic nitrogen (TVB-N) in pork. However, the NIR models lack robustness and often fail when used on a new batch. To handle the problem and obtain better prediction performance, a model updating method based on just-in-time learning (JITL) was proposed in this study. A comprehensive similarity criterion considering both input (spectra) and output (TVB-N content) information was designed. Combining a defined similarity factor, the most relevant samples to new batch samples were selected and a local least square support vector machine model was established in real time based on the selected samples. The results showed that the models updated with JITL approach kept a high predictive performance on new independent batch with prediction error decreasing from 2.95 to 1.60 mg/100 g. The robust models made on selected similar samples combined with JITL model updating strategy can support to make NIR spectroscopy a preferred choice for non-destructive assessment of quality features in pork meat.

Clarification of the Mechanism of Pulse Laser Grinding of Nanosecond Lasers Using High-Speed Camera Imaging

Pulse laser grinding (PLG), as a cutting tool processing method, can not only achieve edge sharpening with high precision, but it can also produce surface modification. For example, polycrystalline cubic boron nitride (PCBN) tools processed by PLG can show increased hardness due to the reduction in defects. However, the mechanism of edge formation under PLG processing remains unclear. In this study, by observing the plasma generated during processing using a high-speed camera, the elementary process for each laser pulse of the PLG process was visualized. The plasma luminescence moved successively through four stages: multipoint luminescence, uniform luminescence, the downward movement of the luminous center, and faint luminescence. By comparing the results of three different laser pulse pitches (0.2, 2, and 20 μm), it was found that the pulse pitch had a significant influence on the PLG processing mode. When the pulse pitch was too small, the sidewall effect was likely to lead to local excess machining. The large pulse pitch resulted in processed surfaces that could not be fully covered by laser irradiation, and it was preferred to remove the decrease threshold subsequently. Thus, the moderate pulse pitch condition showed a superior processed surface compared to the others.

Music for Counting Votes

Music for Counting Votes

Enhancing coding skills with CloudStor SWAN

CloudStor SWAN (AARNet, 2022) is a research-focused web service for running analyses that is available to staff and students at many research institutes and Universities across Australia and New Zealand. In 2021, we used SWAN as a teaching tool in the master-level subject, Computational Genomics (COMP90016) at The University of Melbourne. This subject aims to teach students how to analyse large genomic datasets using best practices software tools, pipelines and student-written, custom code. 
 
 Although CloudStor SWAN was not conceived as a teaching tool, we worked with their technical staff to tailor the service to our use case. This innovative use of existing research infrastructure allowed us to effectively transition the subject to remote learning. Students and staff could log in to the service using their existing University credentials, from anywhere in the world, without the use of a VPN. The ability to access the platform from a web browser allowed for a consistent computing environment for all students regardless of operating system, and without having to worry about software installations on local machines. This presented a significantly improved experience from the custom servers that had been used in the past.
 
 We used SWAN for weekly workshops during semester and for assessment in the form of assignments and an exam. It allowed us to format subject material in Jupyter notebooks where we could seamlessly integrate text, graphics and code. Additionally, assessed code questions can incorporate automatic marking and written submissions can be checked for plagiarism. SWAN also allowed us to introduce students to the UNIX command line, an important skillset that was not previously taught in the University of Melbourne Master of Science (Bioinformatics) program.
 
 From a student perspective, SWAN allowed for a practical skillset to be developed alongside theoretical knowledge from other aspects of the course. The platform was simple to learn and allowed students to focus on the subject content and the tasks asked of them, rather than on the interface. From a teacher’s perspective, having a unified platform allowed for a single set of clear instructions, improved troubleshooting and clearer management of tool versions and software dependencies. The use of Jupyter notebooks simplified lesson plans and assessments by integrating multiple elements into single documents. This element also made the lessons more easily sharable between colleagues and collaborators.
 
 Our integration of this technology into our tertiary teaching has served as a model for a similar use at a different Australian university. We hope to share the lessons learned from this subject, the advantages of using CloudStor SWAN in a teaching environment for both staff and students and provide some advice for others who may want to adapt it to fit their own teaching needs.
 Presentation link: https://youtu.be/8tutCO1hd9c
 References
 
 AARNet. (2022). CloudStor: Access, store, share and work with your data in one place. https://www.aarnet.edu.au/network-and-services/cloud-services/cloudstor