查看更多>>摘要：This research introduces a groundbreaking approach for identification of lung diseases, with a focus on improving patient outcomes and easing the burden on healthcare systems. The primary goal is to enable precise prediction of lung diseases at an early stage. The proposed hybrid model, CNN-GRU-LSTM, is employed for analyzing sequential lung images. An Ensemble Deep Net model is introduced, incorporating CNNs, VGG-based neural networks, and Caps networks. A key innovation is the Spatio Temporal Deep (STD) feature, which captures global and local contextual information across multiple frames. The evaluation of the hybrid model on a diverse dataset demonstrates its effectiveness in identifying subtle patterns indicative of early-stage lung diseases. The Ensemble Deep Net model achieves exceptional accuracy in lesion detection within the complex and variable landscape of lung nodules. The Ensemble Deep Net is leveraged to achieve a remarkable 100% accuracy in classifying diverse lung diseases. This research marks a significant advancement in pulmonary healthcare by providing a valuable tool for early diagnosis and intervention in lung diseases. Also, it holds great promise for enhancing preventive strategies in pulmonary medicine offering a revolutionary path for improving patient care.
查看更多>>摘要：While developing a similar computer model of face recognition, face recognition is a typical assignment for humans and a major component of the human insight framework. To solve the issue of light illumination in this search, the study will apply LDA (Linear Discriminant Analysis) and PCA (Principal Component Analysis) approaches. The modern world has several security challenges. Better biometric security systems, including those that use the retina, finger, signature, voice, iris, and hand, have been developed by security agencies. Recently, face biometrics has been included in this list and has offered an additional method for increasingly secure certification. In intense or interior light patterns, such systems are unable to function as intended. Therefore, it is suggested that face recognition identify and notify the system when a person wanted for investigation is discovered at a certain location while being watched by a CCTV camera. Furthermore, the goal of this search will be to find the light lighting problem's greatest efficiency. When there is light illumination, the system will automatically detect the light density and take appropriate action. The Haar Cascade Classifier technique will be utilized in this search to identify, detect, and extract facial characteristics.
查看更多>>摘要：Today the prediction for seat availability and controlling the flow of passenger in the local trains is a challenging task. The passengers are not aware of time and seat availability which leads them to the waiting list. Also, most of the passengers try to book at the same time and its makes the server to slow down its process. In this paper, the prediction system for ticket availability and calculation of travel time is accomplished. The proposed model is tested with the collection of the several years of data set from the southern railways. The data set possess the details like train number, train name, total passengers, total number of seat, date and time. The proposed prediction system reduces customers effort for easy booking of the train.
查看更多>>摘要：Speech and text is the significant intermediate for human communication. People who have partial vision or blind person can get information from speech. The Braille encoding system represents textual documents in a readable format for the visually challenged persons. This paper projects a speech system model to support visually impaired or blind person in reading the text present on scanned printed text image and also transform recognized text into contracted Braille script. The experimentation of the algorithms was carried out on the text image dataset with different font style and size. Experimental results show that the projected method has a good performance on converting extracting text regions in an image to Grade 2 Braille Script.
Edwar Jacinto G.Holman Montiel A.Fernando Martinez S.
135-142页
查看更多>>摘要：The development of software applications requires the definition of tools and techniques that allow covering and having control over the entire life cycle of the software until it is achieved. This paper proposes an analysis of the types of methodologies that currently exist and proposes the implementation of an easy-to-adapt model such as the guide of foundations for project management, focusing on the use of the four management processes and the ten areas of knowledge as found by the Project Management Institute (PMI). All this in order to obtain a model that defines a relational matrix between software phases and knowledge areas, facilitating the management of projects related to software development.
查看更多>>摘要：Wireless sensor network is one of the emerging fields from past decade for environment sensing and reporting. WSN is expected to be integrated into Internet of Things (IoT) in future trends. Transmission of the aggregated sensed data to Base Station (BS) is termed as routing. This transmission of data is based on limited energy capacity nodes which can neither be charged nor be replaced. So efficient routing scheme is required to be designed which consumes less energy. The sensed data is required to be sent to sink. Sink mobility across the network can be considered as fine strategy to balance the load on the nodes of the network. Proposed scheme virtually divides the network area into four isolated regions (Grids), where each Grid in turn further divided into three tiers based on the distance from the center part of that Grid. After forming the minimum spanning tree among nodes of each Grid, data of each Grid is transmitted to mobile sink node. Proposed scheme proves to have better lifetime with less reconstruction cost than existing techniques.
查看更多>>摘要：To minimize the scheduling issues in parallel computing, we describes the scheduling of independent tasks using makespan with deadline as a scheduling importance. Parallel computing facilities to scheduling on different nodes becomes an important concerns in the ongoing years. In multi-objective heterogeneous computing nodes enhances large amount of scheduling and resource allocation to optimize performance and application tasks are factor that make the scheduling on NP Hard problem. For high scheduling of parallel jobs in grid environment we used metaheuristics to find near optimal solution. In this paper, we study scheduling of parallel jobs and concentrate on high-performance workload and optimizes the parameters such as for instance deadline, makespan and flowtime with the utilization of multi-objective monarch butterfly optimization algorithm (MOMBOA) predicated on multi-objective genetic algorithm (MOGA). Our aim should be to idle for high scheduling of parallel jobs in heterogeneous environment. The end result show the potency of the proposed method MOMBO providing solutions that enhance the performance, makespan and deadline regarding other well-known techniques in the literature.
查看更多>>摘要：The profile-based PWS has situated more effectiveness in making upgrades to the exceptional of net search not too lengthy in the past, with developing utilization of individual and habits skills to profile its customers, which is usually gathered implicitly from question ancient prior, looking history, click-through information, bookmarks, person records, and many others. To shield individual privacy in profile-based PWS, advice a PWS framework often called us that will adaptively generalize profiles through queries at the same time respecting character distinct privacy requisites. It has to bear in mind two contradicting effects for the duration of the hunt approach. First, it makes an try to toughen the hunt excessive-great with the personalization utility of the user profile. Subsequent, it desires to hide the privacy contents current within the person profile to position the privacy hazard beneath manipulate.To aid search high-quality, the quest system which is tailored from a cozy kNN system, after which exhibit how to tremendously toughen search satisfactory with the personalization utility of the person profile. In the comfy kNN scheme, Euclidean distance between an expertise records and a question vector is used to select okay-nearest database documents. To cover the privacy contents, Elliptic Curve Cryptography (ECC) algorithm is used for Encryption and Decryption process. Elliptic curve cryptography (ECC) is an approach to public-key cryptography centered on the algebraic constitution of elliptic curves over finite fields. Elliptic curves are moreover utilized in a quantity of integer factorization algorithms which have functions in cryptography.
NETL
NSTL