Pattern-based Methods Research Articles

Activity-based measures of landscape fragmentation

Abstract Context Landscape fragmentation, which has demonstrated links to habitat loss, increased isolation, a loss of connectivity, and decreased biodiversity, is difficult to quantify. Traditional pattern-based approaches to measuring fragmentation use landscape metrics to quantify aspects of the composition or configuration of landscapes. Objective The objective of this study was to examine the relative improvements of an alternative activity-based approach using the cost of traversing a landscape as a proxy for fragmentation and compare it with the traditional approach. Methods One thousand binary landscapes varying in composition and configuration were simulated, and least-cost path analysis provided the data to calculate the activity-based metrics, which were compared with computed traditional pattern-based metrics. Results Activity-based fragmentation assessments were sensitive to levels of landscape fragmentation, but offered improvements over exiting pattern-based methods in that some metrics varied monotonically across the spectrum of landscape configurations and thus makes their interpretation more holistically meaningful. Conclusions This study provides a modular conceptual framework for assessing fragmentation using activity-based metrics that offer functional improvements over existing pattern-based approaches. While we present a focused theoretical implementation, the process to be measured and the scale of observation can be altered to suit specific user requirements, ecosystems, or species of interest.

Online Trajectory Replanning for Avoiding Moving Obstacles Using Fusion Prediction and Gradient-Based Optimization

In this study, we introduce a novel method for an online trajectory replanning approach for fixed-wing Unmanned Aerial Vehicles (UAVs). Our method integrates moving obstacle predictions within a gradient-based optimization framework. The trajectory is represented by uniformly discretized waypoints, which serve as the optimization variables within the cost function. This cost function incorporates multiple objectives, including obstacle avoidance, kinematic and dynamic feasibility, similarity to the reference trajectory, and trajectory smoothness. To enhance prediction accuracy, we combine physics-based and pattern-based methods for predicting obstacle movements. These predicted movements are then integrated into the online trajectory replanning framework, significantly enhancing the system’s safety. Our approach provides a robust solution for navigating dynamic environments, ensuring both optimal and secure UAV operation.

A Combinatorial Strategy for API Completion: Deep Learning and Heuristics

Remembering software library components and mastering their application programming interfaces (APIs) is a daunting task for programmers, due to the sheer volume of available libraries. API completion tools, which predict subsequent APIs based on code context, are essential for improving development efficiency. Existing API completion techniques, however, face specific weaknesses that limit their performance. Pattern-based code completion methods that rely on statistical information excel in extracting common usage patterns of API sequences. However, they often struggle to capture the semantics of the surrounding code. In contrast, deep-learning-based approaches excel in understanding the semantics of the code but may miss certain common usages that can be easily identified by pattern-based methods. Our insight into overcoming these challenges is based on the complementarity between these two types of approaches. This paper proposes a combinatorial method of API completion that aims to exploit the strengths of both pattern-based and deep-learning-based approaches. The basic idea is to utilize a confidence-based selector to determine which type of approach should be utilized to generate predictions. Pattern-based approaches will only be applied if the frequency of a particular pattern exceeds a pre-defined threshold, while in other cases, deep learning models will be utilized to generate the API completion results. The results showed that our approach dramatically improved the accuracy and mean reciprocal rank (MRR) in large-scale experiments, highlighting its utility.

A bay design problem in less-than-unit-load production warehouse

In this paper, we consider a bay design problem in a less-than-unit-load production warehouse, which is motivated by a real-world problem in a semiconductor company. The objective is to maximize the utilization of the vertical space in bays by considering several practical storage requirements. To solve the problem, a non-linear integer programming model is first formulated. Since the problem is similar to a two-stage cutting stock problem (CSP), a column-and-row generation (CRG) method is developed, in which the original problem is decomposed into a restricted master problem and three subproblems, including two classical column generation subproblems and a row generation subproblem. The two former subproblems are solved as unbounded knapsack problems and for the latter, a two-stage approach is applied. The results of computational experiments on randomly generated instances show that the proposed CRG method is more efficient than the classic column-generation-based method, solving the non-linear model directly and solving a cut model in the literature directly. The results of a case study show that our strategy can improve the utilization of the existing warehouse storage space significantly by about 24%. The CRG method is also tested on basic two-stage two-dimensional CSP benchmarks and its performance is compared to those of other pattern-based methods. The results show its potential for effectively solving the basic two-stage two-dimensional CSPs.

EP-BoxE: A method for hypernym discovery based on extended patterns and box embeddings

Hypernym discovery aims to distinguish potential hypernyms for a query term. However, existing methods for hypernym discovery suffer from the following problems: (1) traditional unsupervised pattern-based methods suffer from low recall; (2) recent supervised box embedding methods are deficient in identifying specific hypernyms. To cope with the above problems, this paper presents a method for hypernym discovery based on Extended Patterns and Box Embeddings (EP-BoxE). Firstly, to acquire more hypernymy relation entity pairs, we identify co-hyponyms of a given term and use their hypernyms as the candidate hypernym set for the given term; Secondly, by analyzing the text corpus, we find that the language patterns also provide additional information for hypernym discovery, which also solves the deficiency of the box embedding methods in identifying specific hypernyms. Finally, experimentations on two domain-specific datasets reveal that EP-BoxE surpasses the performance of popular methods on the majority of evaluation metrics.

Smart Contract Code Repair Recommendation based on Reinforcement Learning and Multi-metric Optimization

A smart contract is a kind of code deployed on the blockchain that executes automatically once an event triggers a clause in the contract. Since smart contracts involve businesses such as asset transfer, they are more vulnerable to attacks, so it is crucial to ensure the security of smart contracts. Because a smart contract cannot be tampered with once deployed on the blockchain, for smart contract developers, it is necessary to fix vulnerabilities before deployment. Compared with many vulnerability detection tools for smart contracts, the amount of automatic fix approaches for smart contracts is relatively limited. These approaches mainly use defined pattern-based methods or heuristic search algorithms for vulnerability repairs. In this paper, we propose RLRep , a reinforcement learning-based approach to provide smart contract repair recommendations for smart contract developers automatically. This approach adopts an agent to provide repair action suggestions based on the vulnerable smart contract without any supervision, which can solve the problem of missing labeled data in machine learning-based repair methods. We evaluate our approach on a dataset containing 853 smart contract programs (programming language: Solidity) with different kinds of vulnerabilities. We split them into training and test set. The result shows that our approach can provide 54.97% correct repair recommendations for smart contracts.

Logical–Mathematical Foundations of a Graph Query Framework for Relational Learning

Relational learning has attracted much attention from the machine learning community in recent years, and many real-world applications have been successfully formulated as relational learning problems. In recent years, several relational learning algorithms have been introduced that follow a pattern-based approach. However, this type of learning model suffers from two fundamental problems: the computational complexity arising from relational queries and the lack of a robust and general framework to serve as the basis for relational learning methods. In this paper, we propose an efficient graph query framework that allows for cyclic queries in polynomial time and is ready to be used in pattern-based learning methods. This solution uses logical predicates instead of graph isomorphisms for query evaluation, reducing complexity and allowing for query refinement through atomic operations. The main differences between our method and other previous pattern-based graph query approaches are the ability to evaluate arbitrary subgraphs instead of nodes or complete graphs, the fact that it is based on mathematical formalization that allows the study of refinements and their complementarity, and the ability to detect cyclic patterns in polynomial time. Application examples show that the proposed framework allows learning relational classifiers to be efficient in generating data with high expressiveness capacities. Specifically, relational decision trees are learned from sets of tagged subnetworks that provide both classifiers and characteristic patterns for the identified classes.

Robust Human Influence across the Troposphere, Surface, and Ocean: A Multivariate Analysis

Abstract Human influence has been robustly detected throughout many parts of the climate system. Pattern-based methods have been used extensively to estimate the strength of model-predicted “fingerprints,” both human and natural, in observational data. However, individual studies using different analysis methods and time periods yield inconsistent estimates of the magnitude of the influence of anthropogenic aerosols, depending on whether they examined the troposphere, surface, or ocean. Reducing the uncertainty of the impact of aerosols on the climate system is crucial for understanding past climate change and obtaining more reliable estimates of climate sensitivity. To reconcile divergent estimates of aerosol effects obtained in previous studies, we apply the same regression-based detection and attribution method to three different variables: mid-to-upper-tropospheric temperature, surface temperature, and ocean heat content. We find that quantitative estimates of human influence in observations are consistent across these three independently monitored components of the climate system. Combining the troposphere, surface, and ocean data into a single multivariate fingerprint results in a small (∼10%) reduction of uncertainty of the magnitude of the greenhouse gas fingerprint, but a large (∼40%) reduction for the anthropogenic aerosol fingerprint. This reduction in uncertainty results in a substantially earlier time of detection of the multivariate aerosol fingerprint when compared to aerosol fingerprint detection time in each of the three individual variables. Our results highlight the benefits of analyzing data across the troposphere, surface, and ocean in detection and attribution studies, and motivate future work to further constrain uncertainties in aerosol effects on climate. Significance Statement Fingerprints of human influence have been detected separately across the troposphere, surface, and ocean. Previous studies examining the different parts of the climate system are difficult to compare quantitatively, however, because they use different methods and cover differ timespans. Here we find consistent estimates of the human influence on the troposphere, surface, and ocean over recent decades when the same fingerprint method and analysis period is used. When we combine the three variables into a single fingerprint, the uncertainty of the influence of anthropogenic aerosols is substantially reduced and the signal is detectable considerably earlier in the observational record. Our results highlight the benefits of performing analysis across different variables instead of focusing on one variable only.

Constructing and Interpreting Causal Knowledge Graphs from News

Many financial jobs rely on news to learn about causal events in the past and present, to make informed decisions and predictions about the future. With the ever-increasing amount of news available online, there is a need to automate the extraction of causal events from unstructured texts. In this work, we propose a methodology to construct causal knowledge graphs (KGs) from news using two steps: (1) Extraction of Causal Relations, and (2) Argument Clustering and Representation into KG. We aim to build graphs that emphasize on recall, precision and interpretability. For extraction, although many earlier works already construct causal KGs from text, most adopt rudimentary pattern-based methods. We close this gap by using the latest BERT-based extraction models alongside pattern-based ones. As a result, we achieved a high recall, while still maintaining a high precision. For clustering, we utilized a topic modelling approach to cluster our arguments, so as to increase the connectivity of our graph. As a result, instead of 15,686 disconnected subgraphs, we were able to obtain 1 connected graph that enables users to infer more causal relationships from. Our final KG effectively captures and conveys causal relationships, validated through experiments, multiple use cases and user feedback.

Feature Selection to Enhance DDoS Detection Using Hybrid N-Gram Heuristic Techniques

Various forms of distributed denial of service (DDoS) assault systems and servers, including traffic overload, request overload, and website breakdowns. Heuristic-based DDoS attack detection is a combination of anomaly-based and pattern-based methods, and it is one of three DDoS attack detection techniques available. The pattern-based method compares a sequence of data packets sent across a computer network using a set of criteria. However, it cannot identify modern assault types, and anomaly-based methods take advantage of the habits that occur in a system. However, this method is difficult to apply because the accuracy is still low, and the false positives are relatively high. Therefore, this study proposes feature selection based on Hybrid N-Gram Heuristic Techniques. The research starts with the conversion process, package extract, and hex payload analysis, focusing on the HTTP protocol. The results show the Hybrid N-Gram Heuristic-based feature selection for the CIC-2017 dataset with the SVM algorithm on the CSDPayload+N-Gram feature with a 4-Gram accuracy rate of 99.86%, MIB- Dataset 2016 with the 2016 algorithm. SVM and CSPayload feature +N-Gram with 100% accuracy for 4-Gram, H2N-Payload Dataset with SVM Algorithm, and CSDPayload+N-Gram feature with 100% accuracy for 4-Gram. As a comparison, the KNN algorithm for 4-Gram has an accuracy rate of 99.44%, and the Neural Network Algorithm has an accuracy rate of 100% for 4-Gram. Thus, the best algorithm for DDoS detection is SVM with Hybrid N-Gram (4-Gram).

Error rate and similarity determination of latent fingerprint chemistry via 1D GC and GC × GC–MS

In response to the National Academy of Science’s report on Strengthening Forensic Science, many researchers have investigated ways of improving and supplementing pattern-based visual methods for latent fingerprint (LF) analysis. Gas chromatography (GC) has been used to study the initial and aged chemistry of latent fingerprints, along with attempting to statistically asses same source association with donor characteristics. Since LFs are composed of a complex mixture of oil, sweat, and other components, sections within the chromatographic profile that have poor resolution could benefit from better separation of components, thus yielding more associations to the correct source. This study investigates the potential of increased source association using comprehensive two-dimensional (GC × GC) and one-dimensional (1D) gas chromatography-mass spectrometry (GC–MS). Three LFs from a sample of thirteen volunteers were collected, derivatized with boron trifluoride–methanol (BF3-MeOH), and analyzed via 1D GC and GC × GC–MS. The statistical strength of source association was performed by evaluating the log likelihood ratios (LLRs) obtained from the intra- and inter-personal pairwise Pearson correlation comparisons (PCCs) and using the normal and kernel density comparison R functions. The 1D GC method provided stronger same-source association due to high intra-personal PCC values; however, lower inter-personal PCCs were acquired from GC × GC. Higher area under the curves (AUCs) and LLR calibration were computed for the 1D GC PCCs. Conversely, poor statistical calibration and discriminating power were obtained for both 1D GC and GC × GC methods when assessing the LLRs despite having high AUCs.

NuKit: A deep learning platform for fast nucleus segmentation of histopathological images.

Nucleus segmentation represents the initial step for histopathological image analysis pipelines, and it remains a challenge in many quantitative analysis methods in terms of accuracy and speed. Recently, deep learning nucleus segmentation methods have demonstrated to outperform previous intensity- or pattern-based methods. However, the heavy computation of deep learning provides impression of lagging response in real time and hampered the adoptability of these models in routine research. We developed and implemented NuKit a deep learning platform, which accelerates nucleus segmentation and provides prompt results to the users. NuKit platform consists of two deep learning models coupled with an interactive graphical user interface (GUI) to provide fast and automatic nucleus segmentation "on the fly". Both deep learning models provide complementary tasks in nucleus segmentation. The whole image segmentation model performs whole image nucleus whereas the click segmentation model supplements the nucleus segmentation with user-driven input to edits the segmented nuclei. We trained the NuKit whole image segmentation model on a large public training data set and tested its performance in seven independent public image data sets. The whole image segmentation model achieves average [Formula: see text] and [Formula: see text]. The outputs could be exported into different file formats, as well as provides seamless integration with other image analysis tools such as QuPath. NuKit can be executed on Windows, Mac, and Linux using personal computers.

Analysis and Forecasting of Monthly Electricity Demand Time Series Using Pattern-Based Statistical Methods

This article provides a solution based on statistical methods (ARIMA, ETS, and Prophet) to predict monthly power demand, which approximates the relationship between historical and future demand patterns. The energy demand time series shows seasonal fluctuation cycles, long-term trends, instability, and random noise. In order to simplify the prediction issue, the monthly load time series is represented by an annual cycle pattern, which unifies the data and filters the trends. A simulation study performed on the monthly electricity load time series for 35 European countries confirmed the high accuracy of the proposed models.

Unsupervised Learning for Feature Selection: A Proposed Solution for Botnet Detection in 5G Networks

The world has seen exponential growth in deploying Internet of Things (IoT) devices. In recent years, connected IoT devices have surpassed the number of connected non-IoT devices. The number of IoT devices continues to grow and they are becoming a critical component of the national infrastructure. IoT devices' characteristics and inherent limitations make them attractive targets for hackers and cyber criminals. Botnet attack is one of the serious threats on the Internet today. This article proposes pattern-based feature selection methods as part of a machine learning (ML)-based botnet detection system. Specifically, two methods are proposed: the first is based on the most dominant pattern feature values and the second is based on maximal frequent itemset mining. The proposed feature selection method uses Gini impurity and an unsupervised clustering method to select the most influential features automatically. The evaluation results show that the proposed methods have improved the performance of the detection system. The developed system has a true positive rate of 100% and a false positive rate of 0% for best performing models. In addition, the proposed methods reduce the computational cost of the system as evidenced by the detection speed of the system.

The vibration knows who you are! A further analysis on usable authentication for smartwatch users

These days, smartwatches are becoming more common and can even operate in stand-alone mode. This increases the need for smartwatches to authenticate users independently without paired smartphones. Currently, password or pattern-based methods can authenticate the smartwatch users in stand-alone mode, but these methods are known to be vulnerable to simple attacks such as shoulder-surfing and password dictionary attacks. In addition, biometric-based methods, which are expected to release on smartwatches in the near future, require inconvenient user interaction or special sensors for measurement. In light of this, we propose a smartwatch user authentication method that does not require any additional sensors or user interaction. Based on the fact that the human body structure affects the way vibrations are absorbed, reflected, and propagated, we designed a smartwatch user authentication method based on a challenge-response structure using vibrations. In our method, a challenge is a set of fresh random vibrations, which are provided by default in current smartwatches, and a response to the challenge is measured by built-in gyroscope and accelerometer sensors. Our earlier study demonstrated that commercial smartwatch users can be authenticated with a low equal error rate (EER) of 1.37 %. In this paper, we extended the analysis of our method on various vibration types by using a prototype setup. As a result, we discovered an outperformed vibration type for user authentication. We conducted further analysis for users with heavier body weights as these individuals are more vulnerable to a not-in-wear attack. Finally, we conducted more advanced impersonation attacks on test participants with one or more similar physical indicators to demonstrate that our method is also secure against a wider range of more complex attacks.

BinVulDet: Detecting vulnerability in binary program via decompiled pseudo code and BiLSTM-attention

Static detection of security vulnerabilities in binary programs is an important research field in software supply chain security. However, existing vulnerability detection methods based on code similarity can only detect known vulnerabilities. Vulnerability features generated by vulnerability pattern-based detection methods are low robust due to the influence of manually defined patterns, compiler diversity, and irrelevant function instructions. In this paper, we propose BinVulDet, which is a binary level vulnerability detection tool for accurate known and unknown vulnerability detection. BinVulDet uses decompilation techniques to obtain pseudo code containing high-level semantic information against the impact of compilation diversity. Then the program slicing technique is used to extract the statements with data dependencies and control dependencies related to the vulnerability. A BiLSTM-attention neural network is used to extract rich contextual semantic information from slice codes to generate more robust vulnerability patterns to detect vulnerabilities. The experimental results show that BinVulDet outperforms the state-of-the-art binary vulnerability detection methods. The FPR and FNR of BinVulDet are 1.04% and 0.89% on average, respectively, which are 3.93% and 22.86% lower than the baseline model on average. BinVulDet can effectively against the influence of compilation diversity and successfully be used for real-world vulnerability detection by being evaluated in three CVE vulnerability projects.

Sentence subjectivity analysis of a political and ideological debate dataset using LSTM and BiLSTM with attention and GRU models

Subjectivity analysis is one of the key tasks in the field of natural language processing. Used to annotate data as subjective or objective, subjectivity analysis can be implemented on its own or as a precursor to other NLP applications such as sentiment analysis, emotion analysis, consumer review analysis, political opinion analysis, document summarization, and question answering systems. The main objective of this article is to test and compare six deep learning methods for subjectivity classification, including Long Short-Term Memory Networks (LSTM), Gated Recurrent Units (GRU), bidirectional GRU, bidirectional LSTM, LSTM with attention, and bidirectional LSTM with attention. We introduced a combination method for subjectivity annotation using lexicon-based and syntactic pattern-based methods. We evaluated the performance of GloVe versus one-hot encoding. We also reformatted, preprocessed, and annotated a political and ideological debate dataset for use in subjectivity analysis. Our research compares favorably with the performance of existing research on subjectivity analysis, achieving very high accuracy and evaluation metrics. LSTM with attention performed the best out of all the methods we tested with an accuracy of 97.39%.

Detecting Outliers in Data Streams Based on Minimum Rare Pattern Mining and Pattern Matching

Outliers are the major factors to influence the accuracy of data-based processing, thus, they must be discovered from collected datasets to guarantee data security. With the widely use of sensors and other monitoring equipment, data streams are becoming the main form of data. However, the huge scale of data streams results in the number of mined rare patterns very large, which makes it hard to effectively detect outliers through pattern-based outlier detection methods. Since minimal rare patterns (MRPs) can represent rare patterns and the number is much smaller, therefore, the use of MRPs on outlier detection can short the time consumption. Based on this idea, this paper proposes an outlier detection approach based on minimal rare pattern, called ODMRP, which is composed of pattern mining phase and pattern matching phase. Specifically, in the pattern mining phase, an improved minimal rare pattern mining algorithm, namely MRPM, is proposed to mine the MRPs from data streams; It first constructs two matrix structures to store information of transactions and frequent 2-patterns, and then apply “pattern extension” operations to extend frequent 2-patterns to longer patterns, at the same time, rare patterns are removed to prevent them participating into “pattern extension” operation to reduce meaningless time cost; In the pattern matching phase, we use the IM-Sunday algorithm to match the mined MRPs with the patterns stored in outlier pattern library, to find potential outliers. Extensive experimental studies show that the proposed ODMRP method can accurately detect outliers from data streams in less overhead.

Deep learning hotspots detection with generative adversarial network-based data augmentation

Lithography process hotspot is a traditional design and quality issue for the integrated circuit manufacturing due to the gap between exposure wavelength and critical feature size. To efficiently detect the hotspot regions and minimize the necessity of conducting expensive lithography simulation experiments, various pattern-based methods have been proposed in the past years. Recent solutions have been focused on implementing deep learning strategies because of the unique strength in imagery classification tasks by employing the artificial neural networks. However, solving the technical bottlenecks such as imbalanced learning, identifying rare hotspots and effective feature extraction remains challenging. For this research, we introduce a hotspot detection method based on a convolutional neural network classifier and enhanced it by the imagery feature extraction and a generative adversarial network data augmentation system. Experimental results show competitive performance compared with the existing works.

Improving sentiment analysis with learning concepts from concept, patterns lexicons and negations

The way of expressing sentiment (−ve/+ve) in the form of textual information depends on the way of thinking of human beings. Identifying aspect extraction and sentiment polarity from written texts is a crucial task. Mainly, a multi-level learning approach for aspect extraction from statistical methods, pattern-based methods, and rule-based methods. This work proposes the application of two probabilistic graphical Latent Dirichlet Allocation (LDA) and Probabilistic Latent Semantic Analysis (PLSA) algorithms to generate latent topic terms as possible aspects. Then frequency-based and Concept lexicons are used to retrieve unigram to multi-word phrases with associated opinion words. Polarity shift is a significant issue that reverses the polarity of the aspects that affect the sentiment classification of the system. Therefore, to improve the performance of the machine learning classification algorithm in ABSA a hybrid approach comprising rule-based methods and a graph-theoretic model is applied to deal with the explicit and implicit polarity shift. The performance of the proposed method is measured using Naive Bayes, a machine learning classification algorithm on two datasets, SemEval 2014 Restaurant and SemEval 2014 Laptop dataset. Experimental result shows that the method for aspect extraction outperforms baseline methods by 86.32% and 82.64% for Restaurant, Laptop dataset, respectively. Similarly, for aspect-based sentiment classification, the accuracy and F1 measure on Restaurant domain 84.73%, 81.28% and 82.06% and 80.71% on the laptop domain.