Search

Software defect prediction is the use of various approaches and procedures to discover and anticipate future flaws in a software project before they become costly and disruptive problems. Organizations may use data analysis, machine learning, and historical defect data to make educated decisions about resource allocation, software testing methodologies, and, ultimately, software product quality. Software defect prediction is the technique of identifying software modules that are likely to have flaws. The suggested system is intended for software defect prediction, combining PCA with a software business analysis model and employing Rank SVM as its primary predictive modeling approach. It includes data collecting from historical defects and relevant data sources, preprocessing to clean and convert the data, feature selection to identify critical indicators, and training a PCA using a software business analysis model utilizing Rank SVM. This methodology, once installed, forecasts the risk of faults in software modules based on their characteristics. The system's output influences resource allocation and testing procedures, hence improving software quality and development efficiency. Continuous monitoring and adjustments assure continuing correctness, making this an efficient option for proactive defect control in software development.

Patent litigation is an expensive and time-consuming legal process. To reduce costs, companies can proactively manage patents using predictive analysis to identify potential plaintiffs, defendants, and patents that may lead to litigation. However, there has been limited progress in predicting patent litigation due to the scarcity of lawsuits, the complexities of intentions, and the diversity of litigation characteristics. To this end, in this paper, we summarize the major causes of patent litigation into multiple aspects: the complex relations among plaintiffs, defendants and patents as well as the diverse content information from them. Along this line, we propose a Multi-aspect Neural Tensor Factorization (MANTF) framework for patent litigation prediction. First, a Pair-wise Tensor Factorization (PTF) module is designed to capture the complex relations among plaintiffs, defendants and patents inherent in a three-dimensional tensor, which will produce factorized latent vectors for companies and patents with pair-wise ranking estimators. Then, to better represent the patents and companies as an aid for PTF, we design a Patent Embedding Network (PEN) module and a Mask Company Embedding Network (MCEN) module to generate content-aware embedding for them, where PEN represents patents based on their meta, textual and graphical features, and MCEN represents companies by integrating their intrinsic features and competitions. Next, to integrate these three modules together, we leverage a Gaussian prior on the difference between factorized representations and content-aware embedding, and train MANTF in an end-to-end way. In the end, final predictions for patent litigation, i.e., the potentially litigated plaintiffs, defendants and patents, can be made with the well-trained model. We conduct extensive experiments on two real-world datasets, whose results prove that MANTF not only helps predict potential patent litigation but also shows robustness under various data sparse situ

Data sparsity is a well-known limitation in the sonar domain. This limitation is a problem when applying data-intensive techniques from the computer vision community, such as deep learning models for detection and classification. One way of extending a sonar dataset is to use simulated sonar images however, these often have the drawback of looking non-realistic when compared to domain data. To overcome the data-sparsity problem as well as for generating realistic-looking sonar images, we introduce a pipeline where the possibilities and limitations of applying cycleGAN to enhance simulated forward-looking sonar images are explored. The results show improved classification performance when training a classifier on enhanced-simulated images compared to training on solely simulated images., Part of proceedings: ISBN 978-1-6654-1821-8QC 20220715

The tag-based recommendation systems that are built based on tensor models commonly suffer from the data sparsity problem. In recent years, various weighted-learning approaches have been proposed to tackle such a problem. The approaches can be categorized by how a weighting scheme is used for exploiting the data sparsity – like employing it to construct a weighted tensor used for weighing the tensor model during the learning process. In this paper, we propose a new weighted-learning approach for exploiting data sparsity in tag-based item recommendation system. We introduce a technique to represent the users’ tag preferences for leveraging the weighted-learning approach. The key idea of the proposed technique comes from the fact that users use different choices of tags to annotate the same item while the same tag may be used to annotate various items in tag-based systems. This points out that users’ tag usage likeliness is different and therefore their tag preferences are also different. We then present three novel weighting schemes that are varied in manners by how the ordinal weighting values are used for labelling the users’ tag preferences. As a result, three weighted tensors are generated based on each scheme. To implement the proposed schemes for generating item recommendations, we develop a novel weighted-learning method called as WRank (Weighted Rank). Our experiments show that considering the users' tag preferences in the tensor-based weightinglearning approach can solve the data sparsity problem as well as improve the quality of recommendation

This thesis addresses the issues of data sparsity in the sonar domain. A data pipeline is set up to generate and enhance sonar data. The possibilities and limitations of using cycleGAN as a tool to enhance simulated sonar images for the purpose of training neural networks for detection and classification is studied. A neural network is trained on the enhanced simulated sonar images and tested on real sonar images to evaluate the quality of these images.The novelty of this work lies in extending previous methods to a more general framework and showing that GAN enhanced simulations work for complex tasks on field data.Using real sonar images to enhance the simulated images, resulted in improved classification compared to a classifier trained on solely simulated images., Denna rapport ämnar undersöka problemet med gles data för djupinlärning i sonardomänen. Ett dataflöde för att generera och höja kvalitén hos simulerad sonardata sätts upp i syfte att skapa en stor uppsättning data för att träna ett neuralt nätverk. Möjligheterna och begränsningarna med att använda cycleGAN för att höja kvalitén hos simulerad sonardata studeras och diskuteras. Ett neuralt nätverk för att upptäcka och klassificera objekt i sonarbilder tränas i syfte att evaluera den förbättrade simulerade sonardatan.Denna rapport bygger vidare på tidigare metoder genom att generalisera dessa och visa att metoden har potential även för komplexa uppgifter baserad på icke trivial data.Genom att träna ett nätverk för klassificering och detektion på simulerade sonarbilder som använder cycleGAN för att höja kvalitén, ökade klassificeringsresultaten markant jämfört med att träna på enbart simulerade bilder.

Deep learning models have been attracting substantial attention in the last few years as they successfully demonstrated remarkable performance on different tasks (e.g., classification and ranking) in various fields. The success of most deep learning models significantly depends on massive training data as these models inherently work by memorizing or distinguishing massive training instances in the data. However, such a huge amount of training data may not be available or accessible all the time due to privacy issues, user experience concerns, or corporation constraints, etc. This leads to the data deficiency issue, which corresponds to the scenarios where we have just a few training instances to accomplish a task. Even though sufficient training data could be acquired, sometimes the data could be very sparse due to the large base of elements (e.g. users and items) in the dataset. This results in the data sparsity issue, which corresponds to the scenarios where we have very limited training instances to pinpoint the characteristics of each element in the data. Both the data deficiency and data sparsity issues limit the expressiveness of strong model capacities of the deep learning models and generally lead to inferior performances on a variety of tasks.In this dissertation, we propose several deep learning frameworks to compensate for the data deficiency and data sparsity in the context of three concrete applications, i.e., customer recommendation in location-based social networks, query recommendation in search engines, and automatic speaker recognition. The methodologies presented in these frameworks span different research areas, including geographical influence modeling on location data, automatic data augmentation via adversarial training, comprehensive instance utilization through metric-learning-based few-shot learning, and knowledge transfer via gradient-based meta-learning. As a result, these methodologies not only tackle specific challenges in the applicat

Recommender systems are widely used and have developed rapidly with the explosion of Web 2.0 technologies. The aim of recommender systems is to provide users with items (products or services) that match the users’ preferences. Recommender systems provide users with personalized online product and service recommendations and are a ubiquitous part of today’s online entertainment smorgasbord. However, many real-world recommender systems suffer from data sparsity and user-preference drift issues which degrade the recommendation performance and lead to a poor user experience. For the user-preference drift issue, time-window and instance decaying approaches are widely applied, but one research gap is that existing methods proposed for adaptation and weighting decay are biased, since the direction of user preference drift was not appropriately addressed in their study. For the data sparsity issue, cross-domain recommender systems are used to handle data sparsity issues. These systems transfer knowledge from one domain that has adequate preference information to another domain that does not. One significant research gap in the existing methods is that they cannot ensure the knowledge extracted from the source domain is consistent with the target domain, which may impact the accuracy of the recommendations. This research addresses the aforementioned research gaps. In this research, to solve these problems and enhance recommender systems, a user profile is enhanced with more information in various domains, including data in different time windows and different categories. For recommender systems with time labels, fuzzy set and fuzzy relation theories are adopted to model uncertain user behavior. A distance measure together with a related statistical guarantee is proposed to detect whether a user preference has drifted or not. A fuzzy user-preference drift detection-based recommendation method is proposed to model user preference and predict user ratings in temporal dynamics.

We address the task of annotating images with semantic tuples. Solving this problem requires an algorithm able to deal with hundreds of classes for each argument of the tuple. In such contexts, data sparsity becomes a key challenge. We propose handling this sparsity by incorporating feature representations of both the inputs (images) and outputs (argument classes) into a factorized log-linear model., Peer Reviewed, Postprint (author's final draft)

Finding similar users in social communities is often challenging, especially in the presence of sparse data or when working with heterogeneous or specialized domains. When computing semantic similarity among users it is desirable to have a measure which allows to compare users w.r.t. any concept in the domain. We propose such a technique which reduces the problems caused by data sparsity, especially in the cold start phase, and enables granular and contextbased adaptive suggestions. It allows referring to a certain set of most similar users in relation to a particular concept when a user needs suggestions about a certain topic (e.g. cultural events) and to a possibly completely different set when the user is interested in another topic (e.g. sport events). Our approach first uses a variation of the spreading activation technique to propagate the users' interests on their corresponding ontology-based user models, and then computes the concept-biased cosine similarity (CBC similarity), a variation of the cosine similarity designed for privileging a particular concept in an ontology. CBC similarity can be used in many adaptation techniques to improve suggestions to users. We include an empirical evaluation on a collaborative filtering algorithm, showing that the CBC similarity works better than the cosine similarity when dealing with sparse data. © Springer International Publishing Switzerland 2013.