scholarly journals Prediction of IoT Traffic Using the Gated Recurrent Unit Neural Network- (GRU-NN-) Based Predictive Model

2021 ◽  
Vol 2021 ◽  
pp. 1-7
Author(s):  
Sonali Appasaheb Patil ◽  
L. Arun Raj ◽  
Bhupesh Kumar Singh
Keyword(s):  
Transfer Learning ◽  
Gradient Boosting ◽  
Traffic Data ◽  
Traffic Characteristics ◽  
Learning Technique ◽  
Long Time ◽  
Statistical Performance Evaluation ◽  
The Comparative Study ◽  
Gated Recurrent Unit ◽  
Evaluation Parameters

Prediction of IoT traffic in the current era has attracted noteworthy attention to utilize the bandwidth and channel capacity optimally. In this paper, the problem of IoT traffic prediction has been studied, and solutions have been proposed by using machine learning method ARIMA and learning time series algorithms such as LSTM and gated recurrent unit (GRU-NN) based on neural networks. The proposed GRU-NN predicts the traffic on the basis of transfer learning. The advantage of the GRU-NN over LSTM is also highlighted by solving the problem of gradient disappearance. The proposed GRU-NN memorizes the traffic characteristics of the IoT environment for a long time which eventually helps the system to forecast the upcoming traffic from the existing traces of the traffic. The proposed GRU-NN makes use of the transfer learning technique to handle the problem of insufficient IoT traffic data along with the gradient boosting training method for achieving better accuracy in predicting the network traffic in the IoT environment. The results reveal that the proposed GRU-NN model outperforms the other traffic predictors in terms of statistical performance evaluation parameters such as MAE, RMSE, MRE, and MSE. The results show that the GRU-NN provides the most accurate predictions followed by the LSTM predictor and then ARIMA and other approaches taken up for the comparative study.

scholarly journals Applying a Deep Q Network for OpenAIs Car Racing Game

2020 ◽  
Author(s):  
Ali Fakhry
Keyword(s):  
Machine Learning ◽  
Reinforcement Learning ◽  
Transfer Learning ◽  
State Of The Art ◽  
Learning Techniques ◽  
Car Racing ◽  
Custom Made ◽  
Learning Technique ◽  
Reward Threshold

The applications of Deep Q-Networks are seen throughout the field of reinforcement learning, a large subsect of machine learning. Using a classic environment from OpenAI, CarRacing-v0, a 2D car racing environment, alongside a custom based modification of the environment, a DQN, Deep Q-Network, was created to solve both the classic and custom environments. The environments are tested using custom made CNN architectures and applying transfer learning from Resnet18. While DQNs were state of the art years ago, using it for CarRacing-v0 appears somewhat unappealing and not as effective as other reinforcement learning techniques. Overall, while the model did train and the agent learned various parts of the environment, attempting to reach the reward threshold for the environment with this reinforcement learning technique seems problematic and difficult as other techniques would be more useful.

Sound Correspondence and the Comparative Study of Miao-Yao Languages (語音對應與苗瑤語比較研究)

2017 ◽  
Vol 10 (1) ◽  
pp. 95-119 ◽  
Author(s):  
Feng Wang (汪鋒) ◽  
Wen Liu (劉文)
Keyword(s):  
Genetic Relationship ◽  
Similar Pattern ◽  
Language Contact ◽  
Historical Linguistics ◽  
Common Ancestor ◽  
Language Family ◽  
Genetic Position ◽  
Long Time ◽  
The Common ◽  
The Comparative Study

Rigorous sound correspondence is fundamental to historical linguistics. It serves as a solid start in studying genetic relationship. Regarding the genetic position of Miao-Yao languages, Li (1937) proposed a hypothesis that the Sino-Tibetan language family consists of Chinese, Tibeto-Burman, Kam-Tai, and Miao-Yao. Benedict (1942; 1975) excluded Miao-Yao from the Sino-Tibetan language family since sound correspondences between Miao-Yao and Chinese were considered to be caused by language contact. The key point in this debate has been ignored for a long time: are the related morphemes proposed in this debate supported by rigorous sound correspondence? In this paper, related morphemes across 11 Miao-Yao languages have been first identified under the requirement of complete sound correspondence, and then analyzed by the Rank Method. The result of the genetic relationship between the 11 Miao-Yao languages has been confirmed. The same procedure has been applied to Sino-Miao-Yao related morphemes, and similar pattern has been found. The Sino-Miao-Yao related morphemes were recognized to be inherited from the common ancestor of Chinese and Miao-Yao. Combined with the result from the perspective of pervasive sound correspondence (Wang 2015), the proposal of a genetic relationship between Chinese and Miao-Yao has been supported. The Inexplicability Principle has been used to weaken the possibility of Sino-Miao-Yao related morphemes being induced by borrowing from Chinese to Miao-Yao, since some sound correspondences are unlikely to be explained by natural phonetic mechanisms. Moreover, related morphemes in Chinese and Miao-Yao have been examined from the perspective of Old Chinese, and such an examination also supports the hypothesis of a genetic relationship between Chinese and Miao-Yao languages. 嚴格的語音對應是歷史比較的基礎，也是判定語源關係的必要條件。在苗瑤語的語源問題研究中，李方桂（1937）提出漢藏語系四語族學說，即漢語、藏緬語、侗台語和苗瑤語。Benedict（1942、1975）則將苗瑤語從漢藏語系中劃分出去，理由是苗瑤語和漢語有對應關係的語素是由接觸造成的。苗瑤語系屬問題的爭議焦點在於苗瑤語和漢語音近義同的一批關係語素是否有嚴格的語音對應支持，然而這一問題一直以來不被重視。本文基於完全對應得到苗瑤語族內部11個語言的關係語素，隨後應用詞階法分析，結果如願所示，這11個語言之間具有發生學關係。同樣的程序應用于漢-苗瑤語關係語素，結果與上述呈現的模式相同，即這些關係語素是來自漢語和苗瑤語共同的祖語，而非語言接觸的產物。結合普遍對應的研究（Wang 2015），漢語和苗瑤語的發生學關係可以得到支持。不可釋原則也顯示漢-苗瑤語關係語素是由苗瑤語從漢語借用的可能性較小，因為二者間的部分語音對應不可能通過自然音變來解釋。此外，從上古漢語的角度對漢-苗瑤語關係語素的校驗也支持二者的同源關係。

scholarly journals The Abnormal Detection for Network Traffic of Power IoT Based on Device Portrait

2020 ◽  
Vol 2020 ◽  
pp. 1-9
Author(s):  
Jiaxuan Fei ◽  
Qigui Yao ◽  
Mingliang Chen ◽  
Xiangqun Wang ◽  
Jie Fan
Keyword(s):  
Machine Learning ◽  
Internet Of Things ◽  
Network Traffic ◽  
Learning Algorithm ◽  
Machine Learning Algorithm ◽  
Traffic Data ◽  
Network Environment ◽  
Security Measures ◽  
Privacy Leakage ◽  
Traffic Characteristics

The construction of power Internet of things is an important development direction for power grid enterprises. Although power Internet of things is a kind of network, it is denser than the ordinary Internet of things points and more complex equipment types, so it has higher requirements for network security protection. At the same time, due to the special information perception and transmission mode in the Internet of things, the information transmitted in the network is easy to be stolen and resold, and traditional security measures can no longer meet the security protection requirements of the new Internet of things devices. To solve the privacy leakage and security attack caused by the illegal intrusion in the network, this paper proposes to construct a device portrait for terminal devices in the power Internet of things and detect abnormal traffic in the network based on device portrait. By collecting traffic data in the network environment, various network traffic characteristics are extracted, and abnormal traffic is analyzed and identified by the machine learning algorithm. By collecting the traffic data in the network environment, the features are extracted from the physical layer, network layer, and application layer of the message, and the device portrait is generated by a machine learning algorithm. According to the established attack mode, the corresponding traffic characteristics are analyzed, and the detection of abnormal traffic is achieved by comparing the attack traffic characteristics with the device portrait. The experimental results show that the accuracy of this method is more than 90%.

Investigation into the Properties of New Materials Having a Possible Effect on the Evaluation Parameters of Sound Insulation and Absorption

2015 ◽  
Vol 220-221 ◽  
pp. 55-59
Author(s):  
Danielius Gužas ◽  
Sergej Anashko ◽  
L. Gogolashvili
Keyword(s):  
Noise Reduction ◽  
Sound Insulation ◽  
New Materials ◽  
Design Quality ◽  
Physical Chemical ◽  
Long Time ◽  
Complex Construction ◽  
The Law ◽  
Evaluation Parameters

The efficiency of creating and designing noise reduction measures greatly depends on physical, chemical, mechanical and other properties of the selected materials. Some of them and sometimes all of those have a big impact on the general sound insulation of designed measures that also include absorption values predetermining their efficiency.As has been known for a long time, the formed heavy partitions and elements acquire increased sound insulation according to the law of mass, i.e. the heavier is the weight per m2 of the surface (kg/km2), the higher is the sound of insulation.To increase the efficiency of noise reduction in barriers, it is necessary to search for some other properties of the material like design, quality, etc.The article describes how calculations and empirical manipulations facilitate finding an optimal complex construction effectively reducing noise.

scholarly journals Identifying Botnet on IoT by Using Supervised Learning Techniques

2019 ◽  
Vol 12 (4) ◽  
pp. 185-193
Author(s):  
Amirhossein Rezaei
Keyword(s):  
Supervised Learning ◽  
Gradient Boosting ◽  
Support Vector ◽  
K Nearest Neighbors ◽  
Malicious Software ◽  
Learning Techniques ◽  
Security Challenges ◽  
Learning Technique ◽  
Security Challenge ◽  
The Moment

The security challenge on IoT (Internet of Things) is one of the hottest and most pertinent topics at the moment especially the several security challenges. The Botnet is one of the security challenges that most impact for several purposes. The network of private computers infected by malicious software and controlled as a group without the knowledge of owners and each of them running one or more bots is called Botnets. Normally, it is used for sending spam, stealing data, and performing DDoS attacks. One of the techniques that been used for detecting the Botnet is the Supervised Learning method. This study will examine several Supervised Learning methods such as; Linear Regression, Logistic Regression, Decision Tree, Naive Bayes, k- Nearest Neighbors, Random Forest, Gradient Boosting Machines, and Support Vector Machine for identifying the Botnet in IoT with the aim of finding which Supervised Learning technique can achieve the highest accuracy and fastest detection as well as with minimizing the dependent variable.

Sign in / Sign up

Export Citation Format

Share Document