Machine learning techniques have been have proven to be more effective than conventional extensively used in the creation of intrusion detection systems (IDS) that can swiftly and automatically identify and classify c...
详细信息
ISBN:
(纸本)9798350348422
Machine learning techniques have been have proven to be more effective than conventional extensively used in the creation of intrusion detection systems (IDS) that can swiftly and automatically identify and classify cyber attacks at the host-and network-levels. A scalable solution is needed since destructive attacks are happening so quickly and are changing all the time. For more investigation, the malware community has access to a number of malware databases. The performance of several machine learning algorithms on a range of datasets that were made available to the general public, however, has not yet been thoroughly evaluated by any study. The publicly available malware datasets should be regularly updated and benchmarked due to the dynamic nature of malware and the continuously evolving attacking techniques. In this study, a deep neural network (DNN), a type of deep learning model, is examined in order to create a flexible and efficient IDS to identify and categorise unexpected and unanticipated cyber threats. In order to analyse a variety of datasets that have been produced throughout time using both static and dynamic methodologies, it is vital to take into account the rapid increase in attacks and the constant evolution of network behaviour. It is simpler to select the most effective algorithm for accurately predicting forthcoming cyber attacks with the help of this type of research. Many publicly available benchmark malware datasets are used to offer a thorough review of DNN and other conventional machine learning classifier studies. The KDDCup 99 dataset and the accompanying hyper parameter selection techniques are used to choose the ideal network parameters and topologies for DNNs. A learning rate of [0.01-0.5] is applied to every 1,000-epoch DNN experiment. A variety of datasets, including NSL-KDD, UNSW-NB15, Kyoto, WSN-DS, and CICIDS 2017, as well as the DNN model that performed well on KDDCup 99 are used to conduct the benchmark. Our DNN model trains a
Human Motion Capture (MoCap) has emerged as the most popular method for human animation production. However, due to joint occlusion, marker shedding, and equipment imprecision, the raw motion data is often corrupted, ...
Human Motion Capture (MoCap) has emerged as the most popular method for human animation production. However, due to joint occlusion, marker shedding, and equipment imprecision, the raw motion data is often corrupted, leading to missing motion data. To address this issue, a missing motion data recovery method utilizing attention-based transformers is proposed in this paper. The proposed model consists of two levels of transformers and a regression head. The first level of transformers extract the spatial features within each frame, and the second level of transformer integrates the per-frame features across time to capture temporal dependencies. The integrated features are then sent to the regression head to derive the complete motion. Extensive experiments on the CMU database demonstrate that the proposed model consistently outperforms the other state-of-the-art methods in recovery accuracy.
Causal discovery in the form of directed acyclic graph (DAG) structure learning finds causal relationships among features of sampled data and has been recognized as one of the most important problems in causal inferen...
详细信息
Causal discovery in the form of directed acyclic graph (DAG) structure learning finds causal relationships among features of sampled data and has been recognized as one of the most important problems in causal inference. Recent works show that DAGs can be learned by solving a continuous optimization problem with a functional equality constraint. However, popular causal models generally assume that the data are i.i.d. in the sense that all data samples are generated by only one underlying causal graph. In this work, we propose a general causal learning model inspired by meta-learning, which aims at finding an invariant DAG over multiple domains and increasing the generalization performance of DAG structure discovery. Mathematically, this model is formulated as a functional constrained bilevel optimization problem that can be solved by our proposed bilevel primal-dual (BPD) algorithm with provable convergence rate guarantees. Numerous numerical experiments demonstrate that the proposed meta-DAG model and BPD algorithm outperform the benchmarks in terms of reconstruction errors and graph Hamming distance.
The term "Cyber-Physical Systems" (CPS) often refers to systems that are both designed and physical, as well as biological. In regard to a CPS, the evolution of physical quantities and distinct software and ...
详细信息
The continuous advancement of remote sensor technology is contributing to a daily surge in data production, necessitating improvements in the accuracy of big data classification. This research proposes a unique featur...
The continuous advancement of remote sensor technology is contributing to a daily surge in data production, necessitating improvements in the accuracy of big data classification. This research proposes a unique feature selection method named as Improved Ant Lion Optimization (IALO) algorithm that integrates the behavior of animal migration with Ant Lion Optimization (ALO) algorithm. The proposed feature selection method enhances the performance of the ALO algorithm. The datasets utilized for the research are iris, wine, Cleveland, and Switzerland datasets. Next, the map reduces framework is utilized and then feature extraction is processed by Linear Discriminant Analysis (LDA). The feature selection is performed by the proposed IALO algorithm and the selected features are given to the Convolutional Neural Network (CNN) classifier for the big data classification. The performance of the proposed method is evaluated by parameters such as accuracy, computation time, specificity, and sensitivity. The proposed method attained a high accuracy of 98.05% with less computation time of 4.8 secs in the wine dataset which is comparatively higher than other existing methods such as Grey Wolf Optimization based CNN (GWO-CNN), Particle Swarm Optimization based CNN (PSO-CNN), Cuckoo Search Optimization based CNN (CSO-CNN), Ant Lion Optimization based CNN (ALO-CNN).
In the recent technologically advanced world, internet serves as a vast repository of data. The exponential growth of online activities has resulted in an abundance of data, making it crucial for businesses to harness...
In the recent technologically advanced world, internet serves as a vast repository of data. The exponential growth of online activities has resulted in an abundance of data, making it crucial for businesses to harness this information to gain meaningful insights for enabling effective decision-making. One significant source of data is social media platforms such as WhatsApp, Twitter, and Facebook. However, the sheer volume of data generated on these platforms presents challenges in terms of data handling and deriving valuable insights. The primary objective of this research work is to analyze social media data and extract valuable insights using advanced analytics techniques. By leveraging social media APIs, data will be collected from various platforms and the clustering algorithms are applied to uncover patterns, trends, and relationships within the dataset. The outcomes of the performed analysis offer valuable insights to drive their strategies towards achieving greater success. By utilizing social media data analytics, organizations gain a deeper understanding of their target audience, identify emerging market trends, and make data-driven decisions. This research aims to assist businesses in improving their strategies, enhancing customer engagement, and achieving better outcomes.
Now a days, without network and advanced computer technologies we can’t able to transfer data among users. Among this the security is the biggest issue with the CPS system. The major issue is figuring out how to incr...
详细信息
Now a days, without network and advanced computer technologies we can’t able to transfer data among users. Among this the security is the biggest issue with the CPS system. The major issue is figuring out how to increase these systems’ security while still doing it effectively. In order to overcome the issues this survey proposes several techniques for providing security of cyber-physical systems using reversible computing and cellular automata. A Cyber-Physical System(CPS) is an intelligent system used for data storing, transferring and protecting. To protect data, many machine(ml) and deep learning(dl) algorithms have been designed, which require more computation, dissipating more *** solve this, we are offering an public key encryption using reversible cellular automata to enhance the security of data in Cyber-Physical Systems
With the development of deep learning and the knowledge graph, artificial intelligence has had a significant influence on the area of education as a result of the rapid growth of this age, which has profoundly altered...
With the development of deep learning and the knowledge graph, artificial intelligence has had a significant influence on the area of education as a result of the rapid growth of this age, which has profoundly altered human productivity and daily life. The university is undergoing a digital transformation. The essence of the knowledge graph is the knowledge base of the semantic network. Using natural language processing (NLP) technology, the university can construct a knowledge graph. Integrating knowledge graphs and deep learning has become one of the most important aspects of further improving the effect of deep learning. The solution to knowledge graph processingdata is “algorithm, totalization, and implementation”. The key technologies of constructing knowledge graphs in multi-dimensional data mainly focus on knowledge ontology definition, knowledge representation, knowledge modeling, knowledge extraction, knowledge fusion, knowledge processing, knowledge computing, and other technologies. The research in this paper shows that the RDF model and algorithm of the knowledge graph have application prospects in multi-dimensional data.
One of the cornerstones of the future generation of computing is cloud computing, a form of internet-based computing. It allows for the instantaneous allocation of Internet-accessible resources whenever they are neede...
详细信息
One of the cornerstones of the future generation of computing is cloud computing, a form of internet-based computing. It allows for the instantaneous allocation of Internet-accessible resources whenever they are needed. Services such as software as a service (SaaS), platform as a service (PaaS), and infrastructure as a service (IaaS) are made available to people and businesses through cloud computing. By giving customers a low-cost, scalable, and financially-independent place to store their data, cloud storage services have seen rapid revenue growth in recent years. An interoperable cloud computing system is built on open architectures and interfaces that support both private and public cloud services. Many businesses today are making the switch to the cloud because of the numerous advantages it offers in terms of data storage, retrieval, management, and availability. While cloud computing is receiving a lot of attention, it is imperative that security problems be addressed first. This study proposes a crypto-stegno mechanism in the cloud as a solution to cloud security problems. This study proposes a practical method for ensuring data privacy and security in cloud storage. In order to obtain their own public and private key, users must first register their information in the cloud and come up with a unique username and password. Cryptography, a means of concealing information and sending information in an unreadable format, is employed in the next level of security. Secure data transmission and storage on the web, electronic commerce, digital media privacy, and ATM transfer are all dependent on cryptography to some degree. Repudiation, integrity, authenticity, and confidentiality are all possible using today's cryptographic methods. Signcryption is a lightweight cryptographic system that combines encryption and signatures in a logical way to achieve higher security with less computing time.
Prompt-based learning improves the performance of Pre-trained Language Models (PLMs) over few-shot learning and is suitable for low-resourced scenarios. However, it is challenging to deploy large PLMs online. Knowledg...
详细信息
Prompt-based learning improves the performance of Pre-trained Language Models (PLMs) over few-shot learning and is suitable for low-resourced scenarios. However, it is challenging to deploy large PLMs online. Knowledge Distillation (KD) can compress large PLMs into small ones; yet, few-shot KD for prompt-tuned PLMs is challenging due to the lack of training data and the capacity gap between teacher and student models. We propose Boost-Distiller, the first few-shot KD algorithm for prompt-tuned PLMs with the help of the out-of-domain data. Apart from distilling the model logits, Boost-Distiller specifically considers heuristically-generated fake logits that improve the generalization abilities of student models. We further leverage the cross-domain model logits, weighted with domain expertise scores that measure the transferablity of out-of-domain instances. Experiments over various datasets show Boost-Distiller consistently outperforms baselines by a large margin.
暂无评论