Machine Learning & Data Mining In Pattern Recognition

Pattern mining derives from the need of discovering hidden knowledge in very large amounts of data, regardless of the form in which it is presented. When it comes to Natural Language Processing (NLP), it arose along the humans' necessity of being understood by computers. In this paper we present an exploratory approach that aims at bringing together the best of both worlds. Our goal is to discover patterns in linguistically processed texts, through the usage of NLP state-of-the-art tools and traditional pattern mining algorithms. Articles from a Portuguese newspaper are the input of a series of tests described in this paper. First, they are processed by an NLP chain, which performs a deep linguistic analysis of text; afterwards, pattern mining algorithms Apriori and GenPrefixSpan are used. Results showed the applicability of pattern mining techniques in textual structured data, and also provided several evidences about the structure of the language.

The problem of sequential pattern mining is one of the several that has deserved particular attention on the general area of data mining. Despite the important developments in the last years, the best algorithm in the area (PrefixSpan) does not deal with gap constraints and consequently doesn't allow for the introduction of background knowledge into the process. In this paper we present the generalization of the PrefixSpan algorithm to deal with gap constraints, using a new method to generate projected databases. Studies on performance and scalability were conducted in synthetic and real-life datasets, and the respective results are presented.

We propose a new cost function for neural network classification: the error density at the origin. This method provides a simple objective function that can be easily plugged in the usual backpropagation algorithm, giving a simple and efficient learning scheme. Experimental work shows the effectiveness and superiority of the proposed method when compared to the usual mean square error criteria in four well known datasets.

Social media platforms have become an important part of our daily lives due to the widespread use of the Internet. They contains a great wealth of valuable information which provide opportunities for us to explore hidden patterns or unknown correlation relationships. The most existing  Association  Rule Mining (ARM)  algorithms  focus  on  mining  associative  patterns  from  large sentiment datasets  based on traditional support-confidence framework. Those algorithms generate a large number of redundant patterns, the majority of which are uninteresting to the user or do not imply a true correlation relationship between related items. In  this  paper,  an  effective  approach  that  integrates  Natural  Languages  Processing (NLP)  techniques, and  ARM concepts in  one  vertical  mining approach in order to discover interesting sentiment associative patterns  is  proposed.  The  proposed  approach  uses  common  NLP  techniques  as  a pre-processing  phase  to  generate  stemmed  wordsets  from  large  sentiment  datasets. In the second phase, the proposed approach pushes the support-all-confidence interestingness measures as a new framework deep during the mining process in order to generate a reduced and complete set of All-Confident Frequent Sentiment Wordsets (ACFSWs) directly from large vertical sentiment datasets.  Furthermore, the generated ACFSWs are used as an underlying knowledge representation for Interesting Sentiment Association Rules (ISARs) discovery. The obtained results show the usefulness and effectiveness of the proposed approach.

Feature extraction process plays an important role in classifying radar target. The extracted features will be fed as the input to the classifier. The incorrect choice of extracted features will cause poor performance of radar classification system. This paper presents the vehicles classification based on combination of different feature extraction algorithm with neural network using forward scattering radar. Hence, the main objective of this paper is to analyze the most suitable feature extraction algorithm which based on different dimensionality reduction technique in order to evaluate the performance of classification system. Three different techniques are used such as the manual and automatic reduction technique (PCA and Z-score) and neural network as classifier to classify the vehicles into their groups either medium or large based on their physical size. The performance of the classification system is determined by the percentage of classification accuracy. The higher percentage of the classification accuracy shows the better performance of the classification system.

Safety is of utmost essence in the aviation sector, both on-ground and in the air. Aviation Turbine kerosene (ATK) commonly referred to as Jet fuel is one of the major resources of the aviation sector, contributing significantly to the operating cost of an airline. Flight safety is a top-notch requirement in air transportation management. Jet fuel quality affects flight safety, and this makes it mandatory to ensure that, at all points in the jet A-1 aviation fuel supply chain, the jet fuel is contamination free and on-spec. Jet fuel quality is determined via various mandatory Joint Inspection Group (JIG) based quality analysis test procedures; both baseline and extensive lab tests by third party labs. Acceptable parameter range has been established for each jet fuel property, the electrical conductivity of jet A-1 fuel must be between 50 to 600 pS/m and the density at 15C must be between 0.775 to 0.840 g/cm 3. Beyond this range, the fuel is deemed off-spec and unsafe for into-plane fuelling operations. This data article presents daily jet fuel test records for jet-A1 fuel. The dataset contains the date of the test, the conductivity, the specific gravity at ambient temperature, the converted specific gravity at 15C, and the temperature of the jet fuel sample under study. All the tests were performed at standard laboratory conditions using approved and certified equipment. The dataset provides an opportunity for developing a predictive model that can be used for jet fuel properties prediction on a given day, based on previous data trends and analysis using data pattern recognition, as an indication of the variation of jet fuel properties with daily weather variation.

Most studies about writer identification are based on English documents and to our knowledge no research has been reported on Farsi or Arabic documents. In this paper, we have proposed a new method for off-line writer identification which is based on Farsi handwriting and text-independent. Based on the idea that has been presented in the previous studies, here we assume handwriting as texture image and a set of features which are based on multi-channel Gabor filters are extracted from preprocessed image of documents. Substantially, the property of proposed method is using of the bank of Gabor filters which is appropriate for structure of Farsi handwritten texts and vision system. Also, a new feature extraction method is proposed which is based on Gabor-energy and moments. For the first, we survey different methods for feature extraction from output of Gabor filters. These methods with co-occurrence matrix and Said method are implemented and experimental results on handwriting of 40 peoples demonstrate that the proposed method achieves better performance on Farsi handwritten documents.

Periodic pattern mining is an emerging technique for knowledge discovery. Most previous approaches have aimed to find only those patterns that exhibit full (or perfect) periodic behavior in databases. Consequently, the existing approaches miss interesting patterns that exhibit partial periodic behavior in a database. With this motivation, this paper proposes a novel model for finding partial periodic patterns that may exist in temporal databases. An efficient pattern-growth algorithm, called Partial Periodic Pattern-growth (3P-growth), is also presented, which can effectively find all desired patterns within a database. Substantial experiments on both real-world and synthetic databases showed that our algorithm is not only efficient in terms of memory and runtime, but is also highly scalable. Finally, the effectiveness of our patterns is demonstrated using two case studies. In the first case study, our model was employed to identify the highly polluted areas in Japan. In the second ...

Association rules are one of the most researched areas of data mining. This is useful in the marketing and retailing strategies. Association mining is to retrieval of a set of attributes shared with a large number of objects in a given database. There are many potential application areas for association rule approach which include design, layout, and customer segregation and so on. The redundancy in association rules affects the quality of the information presented. The goal of redundancy elimination is to improve the quality and usefulness of the rules. Our work aims is to remove hierarchical duplicacy in multi-level, thus reducing the size of the rule set to improve the quality and usefulness without any loss.

Contemporary technologies afford us the opportunity to augment and substitute traditional in-person classes with computer-based resources commonly known as virtual labs. Before the global pandemic, physical classrooms offered a hands-on learning environment for students. However, the pandemic has rendered in-person labs impractical, making it challenging for students to engage with faculty members directly. In the present scenario, the preference for remote education has surged due to safety concerns. With proper internet access, students now have the flexibility to conduct experiments from the comfort of their homes. Despite the inherent challenges associated with online learning, virtual labs serve as a valuable solution, addressing some of these issues. The utilization of simulation-based learning, especially in subjects like physics, chemistry, and biology, contributes significantly to students' comprehension of key concepts. Virtual labs play a pivotal role in fostering problem-solving skills, critical thinking, creativity, imagination, understanding of scientific processes, laboratory skills, as well as heightened interest and perception. Consequently, it becomes imperative for educators to integrate virtual labs into their teaching methodologies to enhance the quality of instruction and improve students' learning outcomes.

In this paper, we delve into the public discourse surrounding COVID-19 on Twitter to unearth the collective sentiments, concerns, and spread of information during the pandemic. By leveraging a dataset of relevant tweets and corresponding ISO country codes, our analysis will map out the geographical and digital landscape of these conversations. The significance of this work lies in its potential to inform public health strategies, shape policymaking, and contribute to social research on crisis communication. Stakeholders ranging from health officials to the public have a vested interest in understanding the contours of this dialogue. Our objective is to craft a data-driven narrative through visualizations that reveal how the world engages with the pandemic on the digital front, providing actionable insights into global and local responses to COVID-19 using Machine Learning techniques.

As a context of biometrics, significant advances have been made in face recognition during the recent decades. Face recognition is one of the most successful applications of image analysis. The accuracy of automated face recognition is greatly affected by varying in lighting between probe and train images. Difference in lighting condition is one of the difficulties in automated face recognition systems. Histogram equalization technique is widely used to diminish the desired effect of different illumination condition between probe and train images by normalizing variation in illumination. Experiments show that normalizing images that has good lighting condition could lead to an increase in recognition error. Wavelet transform, that is well-known as a multiresolution method, is used in features extracting phase. The multiresolution property of wavelet transform is used in extracting feature leading us to have facial feature descriptors at different scales and frequencies. This thesis presents image quality based technique which is measured in terms of luminance to overcome the disadvantage of varying lighting condition to increase the accuracy of face recognition method. 10-fold cross variation is used to investigate the effect of data selection on classification algorithm. At the end, results are compared to investigate the best method for automated face recognition when illumination variation exists.

The recent research of facial expression recognition has made a lot of progress due to the development of deep learning technologies, but some typical challenging problems such as the variety of rich facial expressions and poses are still not resolved. To solve these problems, we develop a new Facial Expression Recognition (FER) framework by involving the facial poses into our image synthesizing and classification process. There are two major novelties in this work. First, we create a new facial expression dataset of more than 200k images with 119 persons, 4 poses and 54 expressions. To our knowledge this is the first dataset to label faces with subtle emotion changes for expression recognition purpose. It is also the first dataset that is large enough to validate the FER task on unbalanced poses, expressions, and zero-shot subject IDs. Second, we propose a facial pose generative adversarial network (FaPE-GAN) to synthesize new facial expression images to augment the data set for tr...

Many video-on-demand (VoD) providers leverage the benefits of cloud computing despite the stemming various security threats. For instance, if the hosted multimedia contents are not encrypted, then the VoD providers must fully trust that the cloud provider will not illegitimately access, distribute, or modify the multimedia contents. Nonetheless, the latest ENISA report showed that approximately 25% of security threats over the last six years have been based on internal misuse. In this paper, we introduce a simple yet effective scheme to protect contents from internal and external threats. Encrypt once-confidentiality, privacy, and integrity (EO-CPI) focuses on protecting data confidentiality, integrity, and users' privacy from untrusted cloud-based VoD systems. We compared EO-CPI with advance encryption standard-Galois counter mode (AES-GCM). Results demonstrated a significant reduction of 79% of the server-side computational overhead when using EO-CPI with a re-encryption degree of one time per five users while preserving the user-side performance.

The global need for effective disease diagnosis remains substantial, given the complexities of various disease mechanisms and diverse patient symptoms. To tackle these challenges, researchers, physicians, and patients are turning to machine learning (ML), an artificial intelligence (AI) discipline, to develop solutions. By leveraging sophisticated ML and AI methods, healthcare stakeholders gain enhanced diagnostic and treatment capabilities. However, there is a scarcity of research focused on ML algorithms for enhancing the accuracy and computational efficiency. This research investigates the capacity of machine learning algorithms to improve the transmission of heart rate data in time series healthcare metrics, concentrating particularly on optimizing accuracy and efficiency. By exploring various ML algorithms used in healthcare applications, the review presents the latest trends and approaches in ML-based disease diagnosis (MLBDD). The factors under consideration include the algorithm utilized, the types of diseases targeted, the data types employed, the applications, and the evaluation metrics. This review aims to shed light on the prospects of ML in healthcare, particularly in disease diagnosis. By analyzing the current literature, the study provides insights into state-of-the-art methodologies and their performance metrics.

Association rules are one of the most researched areas of data mining. This is useful in the marketing and retailing strategies. Association mining is to retrieval of a set of attributes shared with a large number of objects in a given database. There are many potential application areas for association rule approach which include design, layout, and customer segregation and so on. The redundancy in association rules affects the quality of the information presented. The goal of redundancy elimination is to improve the quality and usefulness of the rules. Our work aims is to remove hierarchical duplicacy in multi-level, thus reducing the size of the rule set to improve the quality and usefulness without any loss.

Context: Recent research has used data mining to develop techniques that can guide developers through source code changes. To the best of our knowledge, very few studies have investigated data mining techniques and-or compared their results with other algorithms or a baseline. Objectives: This paper proposes an automatic method for recommending source code changes using four data mining algorithms. We not only use these algorithms to recommend source code changes, but we also conduct an empirical evaluation. Methods: Our investigation includes seven open-source projects from which we extracted source change history at the file level. We used four widely data mining algorithms i.e., Apriori, FP-Growth, Eclat, and Relim to compare the algorithms in terms of performance (Precision, Recall and F-measure) and execution time. Results: Our findings provide empirical evidence that while some Frequent Pattern Mining algorithms, such as Apriori may outperform other algorithms in some cases, the results are not consistent throughout all the software projects, which is more likely due to the nature and characteristics of the studied projects, in particular their change history. Conclusion: Apriori seems appropriate for large-scale projects, whereas

Context: Recent research has used data mining to develop techniques that can guide developers through source code changes. To the best of our knowledge, very few studies have investigated data mining techniques and-or compared their results with other algorithms or a baseline. Objectives: This paper proposes an automatic method for recommending source code changes using four data mining algorithms. We not only use these algorithms to recommend source code changes, but we also conduct an empirical evaluation. Methods: Our investigation includes seven open-source projects from which we extracted source change history at the file level. We used four widely data mining algorithms i.e., Apriori, FP-Growth, Eclat, and Relim to compare the algorithms in terms of performance (Precision, Recall and F-measure) and execution time. Results: Our findings provide empirical evidence that while some Frequent Pattern Mining algorithms, such as Apriori may outperform other algorithms in some cases, the results are not consistent throughout all the software projects, which is more likely due to the nature and characteristics of the studied projects, in particular their change history. Conclusion: Apriori seems appropriate for large-scale projects, whereas

Biometric-based personal verification is a powerful security feature. Biometric systems are used the physiological and/or behavioural characteristics in each individual for verification. Palmprint is a reliable biometric that can be used for identity verification because it is stable and unique for every individual. In this paper, a new approach for personal authentication by palmprint using contourlet transform is presented. Contourlet transform is a multiscale and directional transform that captures image curvatures and smoothness with multidirectional decomposition capability, finely. Our proposed method includes three steps, preprocessing, feature extraction, and classification. In preprocessing stage, the central part of each palmprint is extracted. In feature extraction step, at first, contourlet transform is applied to the central part of palmprint and then features are extracted from created subbands. In this method for each image, 384 features are obtained. Finally, in classification step, naïve Bayes, support vector machine (SVM) and k-nearest neighbour (k-NN) classifiers are employed. Experiments are performed on three databases and recognition accuracies of 99.41%, 92.38%, and 85.34% are obtained on PolyU, COEP and IITD databases using k-NN algorithm, respectively. Experimental results illustrate that our proposed method could be used effectively in personal authentication by palmprint images.

This thesis proposes new approaches to the development of efficient and reliab intrusion detection systems and describes the development of a continuous case-based intrusion detection tool, called AutoGuard. To deal with the uncertainty in the audited data on the target systems, we use Probabilistic and Evidential Reasoning to detect abnormality in the user behavior more effectively. These two methods provide a natural representation of approximate and uncertain information. Evidential Reasoning also provides a formal basis for the key operations of fusion and translation needed to integrate multiple sources of information. Case-based reasoning provides a useful approach for representing knowledge about past intrusions into computer systems and facilitates mechanisms for retrieving and using relevant past cases to solve and reason about new situations. AutoGuard is an advanced case-based reasoning system that analyzes the audit trails of multiuser computer systems in search of impending security violations. AutoGuard presents intrusions as cases within its case-base and uses them to seek out those events within the target system corresponding to known intrusion scenarios. Unlike comparable analysis tools that pattern match sequences of audit records to the expected audit trails of known penetrations, AutoGuard focuses on the class of penetrations and the effects that the individual steps of a penetration have on the system. The case-base is more intuitive to read and update than current penetration rule-bases and allows the system to provide greater functionality to detect impending compromises. v

Minimum support and confidence have been used as criteria for generating association rules in all association rule mining algorithms. These criteria have their natural appeals, such as simplicity; few researchers have suspected the quality of generated rules. In this paper, we examine the rules from a more rigorous point of view by conducting statistical tests. Specifically, we use contingency tables and chi-square test to analyze the data. Experimental results show that one third of the association rules derived based on the support and confidence criteria are not significant, that is, the antecedent and consequent of the rules are not correlated. It indicates that minimum support and minimum confidence do not provide adequate discovery of meaningful associations. The chi-square test can be considered as an enhancement or an alternative solution.

In this paper an intrusion detection algorithm based on GP ensembles is proposed. The algorithm runs on a distributed hybrid multiisland model-based environment to monitor security-related activity within a network. Each island contains a cellular genetic program whose aim is to generate a decision-tree predictor, trained on the local data stored in the node. Every genetic program operates cooperatively, yet independently by the others, by taking advantage of the cellular model to exchange the outmost individuals of the population. After the classifiers are computed, they are collected to form the GP ensemble. Experiments on the KDD Cup 1999 Data show the validity of the approach.

In recent years, Crowd Monitoring techniques have attracted emerging interest in the field of computer vision due to their ability to monitor groups of people in crowded areas, where conventional image processing methods would not suffice. Existing Crowd Monitoring techniques focus heavily on analyzing a crowd as a single entity, usually in terms of their density and movement pattern. While these techniques are well suited for the task of identifying dangerous and emergency situations, they are very limited when it comes to identifying emotion within a crowd. In this work, we propose a novel Crowd Monitoring algorithm based on estimating crowd emotion using Facial Expression Recognition (FER). By isolating different types of emotion within a crowd, we aim to predict the mood of a crowd even in scenes of non-panic. To validate the effectiveness of the proposed algorithm, a series of cross-validation tests are performed using a novel Crowd Emotion dataset with known groundtruth emotions. The results show that the algorithm presented is able to accurately and efficiently predict multiple classes of crowd emotion even in non-panic situations where movement and density information may be incomplete.

Contract Driven Development formalizes functional requirements within component contracts. The process aims to produce higher quality software, reduce quality assurance costs and improve reusability. However, the perceived complexity and cost of requirements formalization has limited the adoption of this approach in industry. In this article, we consider the extent to which the overheads of requirements formalization can be netted off against the reduced quality assurance costs arising from being able to autogenerate functional test interceptors from component contracts. Test-interceptors are used during testing to verify that component contracts are satisfied. In particular, we investigate the impact of contract-driven development on both the quality attributes of the software development process and the quality of the software produced by the process. Empirical data obtained from an actual software project using contract-driven development with test interceptor generation is compa...

In his fascinating paper, "Comparing Timeline Rhythms in Pygmy and Bushmen Music," Adrian Poole offers a new perspective on an old debate regarding the historical significance of the many apparent affinities between African Bushmen and Pygmy music by focusing on "the analysis of one central musical feature that has received little attention in the literature: timeline rhythms." What makes this topic particularly intriguing is the way in which such rhythmic patterns lend themselves to mathematical analysis, a property which enables Poole to produce a very interesting phylogenetic tree from a simple algorithm. His thorough comparative study of these very distinctive rhythms, as found among African hunters isolated for thousands of years in remote regions of that continent with little likelihood of mutual influence, represents a significant contribution to our understanding of certain basic principles underlying both African and African-American rhythms, with additional insights relating to the "deep history" of the timeline on that continent and beyond.

In this paper, we propose a novel feature for palmprint recognition, called Discriminant Orientation Feature (DORIF) based on using Modified Finite Radon Transform (MFRAT) and Two Directional Two Dimensional Linear Discriminant Analysis (2D) 2 LDA. Our proposed method includes two main steps for palmprint feature extraction: (1) Local invariant orientation features are extracted by using MFRAT that handles the palm structure and the variations of illumination and rotation. (2) (2D) 2 LDA is then applied to further remove redundant information and form a discriminant representation more suitable for palmprint recognition. The experimental results for the identification on public database of Hong Kong Polytechnic University (PolyU) demonstrate the effectiveness of the proposed method.

In today's world, data has become a valuable resource for businesses, governments, researchers, and individuals alike. However, to truly extract value from data, it is essential to provide the proper context. Simply collecting and analyzing data without understanding its context can lead to inaccurate conclusions and misguided decision-making. An important factor that drives a successful organization is gathering data that can be analyzed to gain greater insights into the business and enable new opportunities, allowing the business to innovate products/services based on consumer preference. Data is the lifeblood of all businesses, and data-driven decisions can make a significant difference in staying ahead of the competition. Machine learning can be the key to unlocking the value of corporate and customer data, enabling businesses to leverage their data to make more accurate predictions and decisions. It can help businesses to identify patterns and trends in their data that might not be apparent to humans, leading to more accurate predictions and better decisions. Challenge: Machine learning (ML) requires trained professionals and data scientists having good knowledge of programming languages like Python or R. One of the big challenges for data analytics or business people with an SQL background is that it does not matter how good their domain knowledge is, they still cannot contribute much technically when it comes to ML initiatives in their organization. On the other hand, data scientists have good knowledge of statistics, algorithms, Python or R, but they may not have specific data or domain knowledge which can lead to inaccurate predictions and flawed decisions. This paper aims to enable data analysts and data scientists to build and deploy machine learning models on massive datasets using SQL queries without the need for extensive knowledge in programming or data preprocessing.

over the years, a variety of algorithms for finding frequent item sets in very large transaction databases have been developed. The problems of finding frequent item sets are basic in multi level association rule mining, fast algorithms for solving problems are needed. This paper presents an efficient version of apriori algorithm for mining multi-level association rules in large databases to finding maximum frequent itemset at lower level of abstraction. We propose a new, fast and an efficient algorithm (SC-BF Multilevel) with single scan of database for mining complete frequent item sets. To reduce the execution time and increase throughput in new method. Our proposed algorithm works well comparison with general approach of multilevel association rules.

In this paper, we propose a novel feature for palmprint recognition, called Discriminant Orientation Feature (DORIF) based on using Modified Finite Radon Transform (MFRAT) and Two Directional Two Dimensional Linear Discriminant Analysis (2D) 2 LDA. Our proposed method includes two main steps for palmprint feature extraction: (1) Local invariant orientation features are extracted by using MFRAT that handles the palm structure and the variations of illumination and rotation. (2) (2D) 2 LDA is then applied to further remove redundant information and form a discriminant representation more suitable for palmprint recognition. The experimental results for the identification on public database of Hong Kong Polytechnic University (PolyU) demonstrate the effectiveness of the proposed method.

The quick height in innovation has made information in plenty which has both required just as undesirable. So as to truncate the undesirable information and order, plentiful of data mining methods has been actualized. The primary objective of information mining is to gather, pre-process and order the information dependent on specific highlights of soil, for ex, PH, Nitrogen, Oxygen and Carbon percent of soil and foresee the harvests developed for the particular soil type. The application provides a deep insight of the behaviour of the soil in a particular region. It provides a feature extraction method to sort the data set according to the users need and thereby determining the necessary values of certain attributes. The decision makers can easily focus on the content they are interested on under certain context in the sub-classifications by grouping the attributes. Information from different nations are assembled and broke down and were determined based predefined Fuzzy Rules which has returned the precise and recall as result at the end of an output.

Facial affect analysis is perceived as one of themost complex and challenging areas for humanisation ofrobots. Several Facial Expression Recognition (FER)systems apply generic machine learning algorithms toextract facial features. This results in an erroneous classification of previously unseen data. This paper improviseson previous research done on emotion detection and im-plements techniques to leverage the potential of Convolu-tional Neural Networks (CNNs) effectively to classify aset of grayscale images of human faces into seven distinctemotion categories. We experiment with some populartransfer learning models to achieve a maximum accuracyof 98% for the seven-class classification task. To eke outfurther precision and reduce the value loss, we incorpo-rate the Squeeze and Excitation Network to theResNet-50 model, which resulted in a validation accuracyof 99.36%.

• In this example, the minimum support is s min = 3 or σ min = 0.3 = 30%. • There are 2 5 = 32 possible item sets over B = {a, b, c, d, e}. • There are 16 frequent item sets (but only 10 transactions). Christian Borgelt Frequent Pattern Mining 9 (Edge directions are omitted; all edges lead downward.) Christian Borgelt Frequent Pattern Mining 16 Searching for Frequent Item Sets Possible scheme for the search: • Determine the support of the one-element item sets (a.k.a. singletons) and discard the infrequent items / item sets. • Form candidate item sets with two items (both items must be frequent), determine their support, and discard the infrequent item sets. • Form candidate item sets with three items (all contained pairs must be frequent), determine their support, and discard the infrequent item sets. • Continue by forming candidate item sets with four, five etc. items until no candidate item set is frequent. This is the general scheme of the Apriori Algorithm. It is based on two main steps: candidate generation and pruning. All enumeration algorithms are based on these two steps in some form. Christian Borgelt Frequent Pattern Mining 21 • This canonical form will be used to assign unique parents to all item sets. Christian Borgelt Frequent Pattern Mining 32 A Canonical Form for Item Sets • An item set is represented by a code word; each letter represents an item. The code word is a word over the alphabet B, the item base. • There are k! possible code words for an item set of size k, because the items may be listed in any order. • By introducing an (arbitrary, but fixed) order of the items, and by comparing code words lexicographically w.r.t. this order, we can define an order on these code words. Example: abc < bac < bca < cab etc. for the item set {a, b, c} and a < b < c. • The lexicographically smallest (or, alternatively, greatest) code word for an item set is defined to be its canonical code word. Obviously the canonical code word lists the items in the chosen, fixed order. Remark: These explanations may appear obfuscated, since the core idea and the result are very simple. However, the view developed here will help us a lot when we turn to frequent (sub)graph mining. Christian Borgelt Frequent Pattern Mining 33 Canonical Forms and Canonical Parents • Let I be an item set and w c (I) its canonical code word. The canonical parent π c (I) of the item set I is the item set described by the longest proper prefix of the code word w c (I). • Since the canonical code word of an item set lists its items in the chosen order, this definition is equivalent to π c (I) = I − {max i∈I i}. • General Recursive Processing with Canonical Forms: For a given frequent item set I: • Generate all possible extensions J of I by one item (J ⊃ I, |J| = |I| + 1). • Form the canonical code word w c (J) of each extended item set J. • For each J: if the last letter of w c (J) is the item added to I to form J and J is frequent, process J recursively, otherwise discard J. Christian Borgelt Frequent Pattern Mining 34 The Prefix Property • Note that the considered item set coding scheme has the prefix property: The longest proper prefix of the canonical code word of any item set is a canonical code word itself. ⇒ With the longest proper prefix of the canonical code word of an item set I we not only know the canonical parent of I, but also its canonical code word. • Example: Consider the item set I = {a, b, d, e}: • The canonical code word of I is abde. • The longest proper prefix of abde is abd. • The code word abd is the canonical code word of π c (I) = {a, b, d}. • Note that the prefix property immediately implies: Every prefix of a canonical code word is a canonical code word itself. (In the following both statements are called the prefix property, since they are obviously equivalent.) Christian Borgelt Frequent Pattern Mining 35 Searching with the Prefix Property The prefix property allows us to simplify the search scheme: • The general recursive processing scheme with canonical forms requires to construct the canonical code word of each created item set in order to decide whether it has to be processed recursively or not. ⇒ We know the canonical code word of every item set that is processed recursively. • With this code word we know, due to the prefix property, the canonical code words of all child item sets that have to be explored in the recursion with the exception of the last letter (that is, the added item). ⇒ We only have to check whether the code word that results from appending the added item to the given canonical code word is canonical or not. • Advantage: Checking whether a given code word is canonical can be simpler/faster than constructing a canonical code word from scratch. Christian Borgelt Frequent Pattern Mining 36 Christian Borgelt Frequent Pattern Mining 38 Searching with the Prefix Property Exhaustive Search • The prefix property is a necessary condition for ensuring that all canonical code words can be constructed in the search by appending extensions (items) to visited canonical code words. • Suppose the prefix property would not hold. Then: • There exist a canonical code word w and a (proper) prefix v of w, such that v is not a canonical code word. • Forming w by repeatedly appending items must form v first (otherwise the prefix would differ). • When v is constructed in the search, it is discarded, because it is not canonical. • As a consequence, the canonical code word w can never be reached. ⇒ The simplified search scheme can be exhaustive only if the prefix property holds. Christian Borgelt Frequent Pattern Mining 39 Searching with Canonical Forms Straightforward Improvement of the Extension Step: • The considered canonical form lists the items in the chosen item order. ⇒ If the added item succeeds all already present items in the chosen order, the result is in canonical form. ∧ If the added item precedes any of the already present items in the chosen order, the result is not in canonical form. • As a consequence, we have a very simple canonical extension rule (that is, a rule that generates all children and only canonical code words). • Applied to the Apriori algorithm, this means that we generate candidates of size k + 1 by combining two frequent item sets f 1 = {i 1 ,. .. , i k−1 , i k } and f 2 = {i 1 ,. .. , i k−1 , i ′ k } only if i k < i ′ k and ∀j, 1 ≤ j < k : i j < i j+1. Note that it suffices to compare the last letters/items i k and i ′ k if all frequent item sets are represented by canonical code words.

Facial expressions have essential cues to infer the humans state of mind, that conveys adequate information to understand individuals' actual feelings. Thus, automatic facial expression recognition is an interesting and crucial task to interpret the humans cognitive state through the machine. In this paper, we proposed an Exigent Features Preservative Network (EXPERTNet), to describe the features of the facial expressions. The EXPERTNet extracts only pertinent features and neglect others by using exigent feature (ExFeat) block, mainly comprises of elective layer. Specifically, elective layer selects the desired edge variation features from the previous layer outcomes, which are generated by applying different sized filters as 1  1, 3  3, 5  5 and 7  7. Different sized filters aid to elicits both micro and high-level features that enhance the learnability of neurons. ExFeat block preserves the spatial structural information of the facial expression, which allows to discriminate between different classes of facial expressions. Visual representation of the proposed method over different facial expressions shows the learning capability of the neurons of different layers. Experimental and comparative analysis results over four comprehensive datasets: CK+, MMI DISFA and GEMEP-FERA, ensures the better performance of the proposed network as compared to existing networks.

In this paper the usage of convolution neural networks has been considered for solving the problem of emotion recognition by images with facial expression. Emotion recognition is a complex task and the result of recognition is highly dependent on the choice of the neural network architecture. In this paper various architectures of convolutional neural networks were reviewed and training experiments were conducted on selected neural networks. The overviewed neural network architectures were trained on FER2013 and AffectNet datasets, both widely used for emotion recognition experiments. A comparison of the selected neural network architectures was made using the accuracy metrics. At the end of this paper the comparative analysis was made and obtained results were overviewed.

Facial expressions have essential cues to infer the humans state of mind, that conveys adequate information to understand individuals' actual feelings. Thus, automatic facial expression recognition is an interesting and crucial task to interpret the humans cognitive state through the machine. In this paper, we proposed an Exigent Features Preservative Network (EXPERTNet), to describe the features of the facial expressions. The EXPERTNet extracts only pertinent features and neglect others by using exigent feature (ExFeat) block, mainly comprises of elective layer. Specifically, elective layer selects the desired edge variation features from the previous layer outcomes, which are generated by applying different sized filters as 1  1, 3  3, 5  5 and 7  7. Different sized filters aid to elicits both micro and high-level features that enhance the learnability of neurons. ExFeat block preserves the spatial structural information of the facial expression, which allows to discriminate between different classes of facial expressions. Visual representation of the proposed method over different facial expressions shows the learning capability of the neurons of different layers. Experimental and comparative analysis results over four comprehensive datasets: CK+, MMI DISFA and GEMEP-FERA, ensures the better performance of the proposed network as compared to existing networks.

Convolution Neural Network is one of the phenomenal formulations in the field of pattern recognition research, computer vision and image processing. It helped to facilitate many theories into real working models. One of them being Facial expression recognition (FER) which has been benefited with the development of CNN architectures, especially the ResNet architecture which has emerged as the winner of ImageNet 2015 competition. Residual Neural Network (ResNet) inculcates the idea of skip connections and became the most cited neural network. In this paper we have analyzed an extensive view of addressing facial expression and emotion recognition with the assistance of ResNet. We have particularly emphasized upon Resnet-18, Resnet-50, SobelResNet, ResNet with Attention mechanism and deformable convolution, Emotion recognition using heart rate variability analysis and ResNet, 3D inception ResNet layers.

Due to the voluminous amount of itemsets that are generated, the association rules extracted from these itemsets contain redundancy, and designing an effective approach to address this issue is of paramount importance. Although multiple algorithms were proposed in recent years for mining closed association rules most of them underperform in terms of run time or memory. Another issue that remains challenging is the nature of the dataset. While some of the existing algorithms perform well on dense datasets others perform well on sparse datasets. This paper aims to handle these drawbacks by using a genetic algorithm for mining closed association rules. Recent studies have shown that genetic algorithms perform better than conventional algorithms due to their bitwise operations of crossover and mutation. Bitwise operations are predominantly faster than conventional approaches and bits consume lesser memory thereby improving the overall performance of the algorithm. To address the redundancy in the mined association rules a treebased pruning algorithm has been designed here. This works on the principle of minimal antecedent and maximal consequent. Experiments have shown that the proposed approach works well on both dense and sparse datasets while surpassing existing techniques with regard to run time and memory.

Facial expression recognition in a wild situation is a challenging problem in computer vision research due to different circumstances, such as pose dissimilarity, age, lighting conditions, occlusions, etc. Numerous methods, such as point tracking, piecewise affine transformation, compact Euclidean space, modified local directional pattern, and dictionary-based component separation have been applied to solve this problem. In this paper, we have proposed a deep learning-based automatic wild facial expression recognition system where we have implemented an incremental active learning framework using the VGG16 model developed by the Visual Geometry Group. We have gathered a large amount of unlabeled facial expression data from Intelligent Technology Lab (ITLab) members at Inha University, Republic of Korea, to train our incremental active learning framework. We have collected these data under five different lighting conditions: good lighting, average lighting, close to the camera, far from the camera, and natural lighting and with seven facial expressions: happy, disgusted, sad, angry, surprised, fear, and neutral. Our facial recognition framework has been adapted from a multi-task cascaded convolutional network detector. Repeating the entire process helps obtain better performance. Our experimental results have demonstrated that incremental active learning improves the starting baseline accuracy from 63% to average 88% on ITLab dataset on wild environment. We also present extensive results on face expression benchmark such as Extended Cohn-Kanade Dataset, as well as ITLab face dataset captured in wild environment and obtained better performance than state-of-the-art approaches.

The paper includes research on discovering new methods for multilingual speech recognition and comparing the effectiveness of the existing solutions with the proposed novelty approaches. The audio and textual multilingual dataset contains multilingual sentences where each sentence contains words from two different languages - English and Kannada. Our proposed speech recognition process includes preprocessing and splitting each audio sentence based on words, which is then given as input to the DL translator (using MFCC features) along with next word predictions. The use of a Next Word Prediction model along with the DL translator to accurately identify the words and convert to text. Similarly the other approach proposed is the use of cosine similarity where the speech recognition is based on the similarity between word uttered and the generated training dataset. Our models were trained on an audio and textual dataset that were generated by the team members and the test accuracies wer...

We describe an unsupervised host-based intrusion detection system based on system calls arguments and sequences. We define a set of anomaly detection models for the individual parameters of the call. We then describe a clustering process which helps to better fit models to system call arguments, and creates interrelations among different arguments of a system call. Finally, we add a behavioral Markov model in order to capture time correlations and abnormal behaviors. The whole system needs no prior knowledge input; it has a good signal to noise ratio, and it is also able to correctly contextualize alarms, giving the user more information to understand whether a true or false positive happened, and to detect global variations over the entire execution flow, as opposed to punctual ones over individual instances.

In computer vision and image processing, object detection algorithms are used to detect semantic objects of certain classes of images and videos. Object detector algorithms use deep learning networks to classify detected regions. Unprecedented advancements in Convolutional Neural Networks (CNN) have led to new possibilities and implementations for object detectors. An object detector which uses a deep learning algorithm detect objects through proposed regions, and then classifies the region using a CNN. Object detectors are computationally efficient unlike a typical CNN which is computationally complex and expensive. Object detectors are widely used for face detection, recognition, and object tracking. In this thesis, deep learning based object detection algorithms are implemented to classify facially expressed emotions in real-time captured through a webcam. A typical CNN would classify images without specifying regions within an image, which could be considered as a limitation towards better understanding the network performance which depend on different training options. It would also be more difficult to verify whether a network have converged and is able to generalize, which is the ability to classify unseen data, data which was not part of the training set. Fast Region-based Convolutional Neural Network, an object detection algorithm; used to detect facially expressed emotion in real-time by classifying proposed regions. The Fast R-CNN is trained using a high-quality video database, consisting of 24 actors, facially expressing eight different emotions, obtained from images which were processed from 60 videos per actor. An object detector's performance is measured using various metrics. Regardless of how an object detector performed with respect to average precision or miss rate, doing well on such metrics would not necessarily mean that the network is correctly classifying regions. This may result from the fact that the network model has been over-trained. In our work we showed that object detector algorithm such as Fast R-CNN performed surprisingly well in classifying facially expressed emotions in real-time, performing better than CNN.

This paper proposes a kNN model-based feature selection method aimed at improving the efficiency and effectiveness of the ReliefF method by: (1) using a kNN model as the starter selection, aimed at choosing a set of more meaningful representatives to replace the original data for feature selection; (2) integration of the Heterogeneous Value Difference Metric to handle heterogeneous applications-those with both ordinal and nominal features; and (3) presenting a simple method of difference function calculation based on inductive information in each representative obtained by kNN model. We have evaluated the performance of the proposed kNN model-based feature selection method on toxicity dataset Phenols with two different endpoints. Experimental results indicate that the proposed feature selection method has a significant improvement in the classification accuracy for the trial dataset.

This paper proposes a kNN model-based feature selection method aimed at improving the efficiency and effectiveness of the ReliefF method by: (1) using a kNN model as the starter selection, aimed at choosing a set of more meaningful representatives to replace the original data for feature selection; (2) integration of the Heterogeneous Value Difference Metric to handle heterogeneous applications-those with both ordinal and nominal features; and (3) presenting a simple method of difference function calculation based on inductive information in each representative obtained by kNN model. We have evaluated the performance of the proposed kNN model-based feature selection method on toxicity dataset Phenols with two different endpoints. Experimental results indicate that the proposed feature selection method has a significant improvement in the classification accuracy for the trial dataset.

In this paper we address confidentiality issues in distributed data clustering, particularly the inference problem. We present KDEC-S algorithm for distributed data clustering, which is shown to provide mining results while preserving confidentiality of original data. We also present a confidentiality framework with which we can state the confidentiality level of KDEC-S. The underlying idea of KDEC-S is to use an approximation of density estimation such that the original data cannot be reconstructed to a given extent.

Minimum support and confidence have been used as criteria for generating association rules in all association rule mining algorithms. These criteria have their natural appeals, such as simplicity; few researchers have suspected the quality of generated rules. In this paper, we examine the rules from a more rigorous point of view by conducting statistical tests. Specifically, we use contingency tables and chi-square test to analyze the data. Experimental results show that one third of the association rules derived based on the support and confidence criteria are not significant, that is, the antecedent and consequent of the rules are not correlated. It indicates that minimum support and minimum confidence do not provide adequate discovery of meaningful associations. The chi-square test can be considered as an enhancement or an alternative solution.

Mining generalized association rules among items in the presence of taxonomy and with nonuniform minimum support has been recognized as an important model in data mining. In our previous work, we have investigated this problem and proposed two algorithms, MMS_Cumulate and MMS_Stratify. In real applications, however, the work of discovering interesting association rules is an iterative process. The analysts need to repeatedly adjust the constraint of minimum support and/or minimum confidence to discover real informative rules. How to reduce the response time for each remining process thus becomes a crucial issue. In this paper, we examined the problem of maintaining the discovered multi-supported generalized association rules when the multiple minimum support constraint is updated. Empirical evaluation showed that the proposed RM_MMS algorithm is very efficient and has good linear scale-up characteristic.

Safety is of utmost essence in the aviation sector, both on-ground and in the air. Aviation Turbine kerosene (ATK) commonly referred to as Jet fuel is one of the major resources of the aviation sector, contributing significantly to the operating cost of an airline. Flight safety is a top-notch requirement in air transportation management. Jet fuel quality affects flight safety, and this makes it mandatory to ensure that, at all points in the jet A-1 aviation fuel supply chain, the jet fuel is contamination free and on-spec. Jet fuel quality is determined via various mandatory Joint Inspection Group (JIG) based quality analysis test procedures; both baseline and extensive lab tests by third party labs. Acceptable parameter range has been established for each jet fuel property, the electrical conductivity of jet A-1 fuel must be between 50 to 600 pS/m and the density at 15C must be between 0.775 to 0.840 g/cm 3. Beyond this range, the fuel is deemed off-spec and unsafe for into-plane fuelling operations. This data article presents daily jet fuel test records for jet-A1 fuel. The dataset contains the date of the test, the conductivity, the specific gravity at ambient temperature, the converted specific gravity at 15C, and the temperature of the jet fuel sample under study. All the tests were performed at standard laboratory conditions using approved and certified equipment. The dataset provides an opportunity for developing a predictive model that can be used for jet fuel properties prediction on a given day, based on previous data trends and analysis using data pattern recognition, as an indication of the variation of jet fuel properties with daily weather variation.

Digital libraries and intranets often include a wide mix of document types, ranging from cat- alog entries consisting entirely of metadata, to documents with both substantial text content and thorough metadata tagging, to documents with hardly any metadata at all. This paper ...