004 Datenverarbeitung; Informatik
Refine
Document Type
Institute
- Fachbereich Ingenieur- und Naturwissenschaften (7) (remove)
Has Fulltext
- yes (7)
Is part of the Bibliography
- yes (7)
Keywords
- KNN (1)
- RNN (1)
- SVM (1)
- data preprocessing (1)
- epithelium (1)
- inflammatory change (1)
- machine learning (1)
- mask (1)
- microscopic image (1)
- neural network (1)
The article discusses the approach to solving the problem of reducing time spent on the preparation of medical images for teaching neural networks, by reducing the time of creating masks for images. The task is considered on the example of processing images of the mucous membrane of the paranasal sinus. The specifics of the task did not allow effectively using existing software solutions. During the study, a software solution was proposed, which made it possible to radically reduce the time of creating masks for images. The article also analyzes the shortcomings of the automated creation of masks, as well as the directions of their solution. The loss of time due to the adjustment of the color palette can be reduced even more to 1-2 minutes, the average deviation is 7.61%.
The object of the study is the process of identifying the state of a computer network. The subject of the study are the methods of identifying the state of computer networks. The purpose of the paper is to improve the efficacy of intrusion detection in computer networks by developing a method based on transformer models. The results obtained. The work analyzes traditional machine learning algorithms, deep learning methods and considers the advantages of using transformer models. A method for detecting intrusions in computer networks is proposed. This method differs from known approaches by utilizing the Vision Transformer for Small-size Datasets (ViTSD) deep learning algorithm. The method incorporates procedures to reduce the correlation of input data and transform data into a specific format required for model operations. The developed methods are implemented using Python and the GOOGLE COLAB cloud service with Jupyter Notebook. Conclusions. Experiments confirmed the efficiency of the proposed method. The use of the developed method based on the ViTSD algorithm and the data preprocessing procedure increases the model's accuracy to 98.7%. This makes it possible to recommend it for practical use, in order to improve the accuracy of identifying the state of a computer system.
European Union (EU) member states consider themselves bulwarks of democracy and freedom of speech. However, there is a lack of empirical studies assessing possible violations of these principles in the EU through Internet censorship. This work starts addressing this research gap by investigating Internet censorship in Spain over 2016-2020, including the controversial 2017 Catalan independence referendum. We focus, in particular, on network interference disrupting the regular operation of Internet services or contents. We analyzed the data collected by the Open Observatory of Network Interference (OONI) network measurement tool. The measurements targeted civil rights defending websites, secure communication tools, extremist political content, and information portals for the Catalan referendum. Our analysis indicates the existence of advanced network interference techniques that grow in sophistication over time. Internet Service Providers (ISPs) initially introduced information controls for a clearly defined legal scope (i.e., copyright infringement). Our research observed that such information controls had been re-purposed (e.g., to target websites supporting the referendum). We present evidence of network interference from all the major ISPs in Spain, serving 91% of mobile and 98% of broadband users and several governmental and law enforcement authorities. In these measurements, we detected 16 unique blockpages, 2 Deep Packet Inspection (DPI) vendors, and 78 blocked websites. We also contribute an enhanced domain testing methodology to detect certain kinds of Transport Layer Security (TLS) blocking that OONI could not initially detect. In light of our experience analyzing this dataset, we also make suggestions on improving the collection of evidence of network interference.
Das Erkennen von Personen auf Fotos und in Videos geschieht in aktuellen Klassifizierungsverfahren durch die Zuordnung eines Köperteils zu jedem Pixel. Anschließend werden die gefundenen Körperteile zu Skeletten zusammengefasst. Im Falle mehrerer abgebildeter Personen ergibt sich das Problem der Zuordnung der Körperteile zu den verschiedenen Skeletten. Es wurde in dieser Arbeit ein auf dem Suurballe-Algorithmus basierendes graphentheoretisches Verfahren entwickelt, das diese Aufgabe löst. Aufbauend auf im Wesentlichen abstandsabhängigen Kantengewichten wird eine k-kürzeste-disjunkte-Wege-Suche durchgeführt. Unter Einbezug von möglicherweise fehlenden Körperteilen durch Unsichtbar-Knoten und automatisches Aussortieren zu viel gefundener Körperteile werden so die k-optimalen Skelette gefunden. Die Methode wurde an idealisierten computergenerierten Bildern mit einer Trefferquote von 100 % gefundener Personen getestet. Ein Test mit realen Bilddaten lieferte eine Trefferquote von 71,7 %.
In recent years, 3D movies and streaming films have become increasingly popular. Even mobile devices, such as mobile phones and tablet computers, are becoming more popular, more powerful and have better multimedia capabilities. Nevertheless, compact mobile devices to broadcast live 3D videos in real time are barely available. In this article, we provide a modular and mobile solution that allows 3D video streaming in real time at 25 frames per second and with a resolution of 1280 × 720 pixels (720p). As operating system, we use standard hardware components combined with Android. Furthermore, we will describe the restraints of the development, and how they were solved.
This paper introduces procedures, which facilitate the development of as well as the linking between modular data processing processes into different and versatile system solutions. Based on the concept of data flow diagrams (Yourdon 1989), processes and data flows are used. With the ability of a process to harness any amount of in- and outputs with different sources and destinations, it is possible to model any conceivable data processing problem. By aggregating a diagram as a new process, it becomes possible to outsource partial process chains to different abstraction levels. As a result, even very complex solutions can be modeled and explained in a clear way. The introduced uniform termination flow while traversing the graph creates a good readability, which also makes it possible to quickly detect weaknesses. Moreover, this solution is comprehensive and advances the cooperation in teams and development.
A domain specific language for the automatic generation of parsers classes for text protocols
(2014)
ABNF is a language for the definition of the formal syntax of technical specifications and is widely used for the definition of textual protocol messages of many internet protocols. The automatic generation of parser classes for ABNF specifications is currently very limited, because ABNF only defines the transfer syntax of the text messages and does not define names for the set of production rules. The lack of name definitions within ABNF rules does not allow to automatically generate expressiveness and meaningful program code for ABNF specifications. In this paper we present X-ABNF, which is a domain-specific language (DSL) for the definition of name-bindings for ABNF rules. The name-bindings with X-ABNF facilitates to generate a concise and meaningful code for an ABNF specification. Additionally, we show that the name-binding can also be used for language extensions through macro programming to dynamically access ABNF encoded text data within source code. We have used Xtext and Xtend for the implementation of the language grammar of ABNF and X-ABNF, which provide good tool support and code generation capabilities.