Artificial Intelligence

1903 Submissions

[39] viXra:1903.0412 [pdf] submitted on 2019-03-22 08:46:46

Faster and Simpler Deep Learning

Authors: George Rajna
Comments: 44 Pages.

Artificial intelligence systems based on deep learning are changing the electronic devices that surround us. [26] A team of researchers affiliated with several institutions in Germany and the U.S. has developed a deep learning algorithm that can be used for motion capture of animals of any kind. [25] In 2016, when we inaugurated our new IBM Research lab in Johannesburg, we took on this challenge and are reporting our first promising results at Health Day at the KDD Data Science Conference in London this month. [24] The research group took advantage of a system at SLAC's Stanford Synchrotron Radiation Lightsource (SSRL) that combines machine learning-a form of artificial intelligence where computer algorithms glean knowledge from enormous amounts of data-with experiments that quickly make and screen hundreds of sample materials at a time. [23] Researchers at the UCLA Samueli School of Engineering have demonstrated that deep learning, a powerful form of artificial intelligence, can discern and enhance microscopic details in photos taken by smartphones. [22] Such are the big questions behind one of the new projects underway at the MIT-IBM Watson AI Laboratory, a collaboration for research on the frontiers of artificial intelligence. [21]
Category: Artificial Intelligence

[38] viXra:1903.0403 [pdf] submitted on 2019-03-21 08:25:56

Machine Learning World's Oceans

Authors: George Rajna
Comments: 45 Pages.

Oceanographers studying the physics of the global ocean have long found themselves facing a conundrum: Fluid dynamical balances can vary greatly from point to point, rendering it difficult to make global generalizations. [26] The analysis of sensor data of machines, plants or buildings makes it possible to detect anomalous states early and thus to avoid further damage. [25] Physicists in the US have used machine learning to determine the phase diagram of a system of 12 idealized quantum particles to a higher precision than ever before. [24] The research group took advantage of a system at SLAC's Stanford Synchrotron Radiation Lightsource (SSRL) that combines machine learning-a form of artificial intelligence where computer algorithms glean knowledge from enormous amounts of data-with experiments that quickly make and screen hundreds of sample materials at a time. [23] Researchers at the UCLA Samueli School of Engineering have demonstrated that deep learning, a powerful form of artificial intelligence, can discern and enhance microscopic details in photos taken by smartphones. [22] Such are the big questions behind one of the new projects underway at the MIT-IBM Watson AI Laboratory, a collaboration for research on the frontiers of artificial intelligence. [21]
Category: Artificial Intelligence

[37] viXra:1903.0388 [pdf] submitted on 2019-03-22 03:27:14

Comments on the Book \architects of Intelligence" by Martin Ford in the Light of the SP Theory of Intelligence

Authors: J Gerard Wolff
Comments: 53 Pages.

The book "Architects of Intelligence" by Martin Ford presents conversations about AI between the author and influential researchers. Issues considered in the book are described in relation to features of the "SP System", meaning the "SP Theory of Intelligence" and its realisation in the "SP Computer Model", both outlined in an appendix. The SP System has the potential to solve most of the problems in AI described in the book, and some others. Strengths and potential of the SP System, which in many cases contrast with weaknesses of deep neural networks (DNNs), include the following: a top-down research strategy has yielded a system with a favourable combination of conceptual "Simplicity" with descriptive or explanatory "Power"; the SP System has strengths and potential with both symbolic and non-symbolic kinds of knowledge and processing; the system has strengths and long-term potential in pattern recognition; it is free of the tendency of DNNs to make large and unexpected errors in recognition; the system has strengths and potential in unsupervised learning, including grammatical inference; the SP Theory of Intelligence provides a theoretically coherent basis for generalisation and the avoidance of under- or over-generalisations; that theory of generalisation may help improve the safety of driverless; the SP system, unlike DNNs, can achieve learning from a single occurrence or experience; it has relatively tiny demands for computational resources and volumes of data, with potential for much higher speeds in learning; the system, unlike most DNNs, has strengths in transfer learning; unlike DNNs, it provides transparency in the representation of knowledge and an audit trail for all its processing; the system has strengths and potential in the processing of natural language; it exhibits several different kinds of probabilistic reasoning; the system has strengths and potential in commonsense reasoning and the representation of commonsense knowledge; other strengths include information compression, biological validity, scope for adaptation, and freedom from catastrophic forgetting. Despite the importance of motivations and emotions, no attempt has been made in the SP research to investigate these areas.
Category: Artificial Intelligence

[36] viXra:1903.0370 [pdf] submitted on 2019-03-21 01:14:08

[coqtp – Q*cert Ocaml Python Gan] as an Informatics & Computing Platform in the Context of Cryo-em Image Processing & Big Data Research.

Authors: Nirmal Tej Kumar
Comments: 4 Pages. Short Communication & Technical Notes

[CoqTP – q*cert- Ocaml- Python - GAN] as an Informatics & Computing Platform in the Context of cryo-EM Image Processing & BIG DATA Research. Importance of GAN & Theorem Provers For Better cryo-EM Image Processing/IoT/HPC.
Category: Artificial Intelligence

[35] viXra:1903.0341 [pdf] submitted on 2019-03-18 09:46:04

Organic Principal Component Analysis

Authors: George Rajna
Comments: 41 Pages.

Researchers at A*STAR have compared six data-analysis processes and come up with a clear winner in terms of speed, quality of analysis and reliability. [25] Researchers at Max Planck Institute for the Science of Light and Friedrich Alexander University in Erlangen, Germany have recently demonstrated that a molecule can be turned into a coherent two-level quantum system. [24] Researchers at the University of Dundee have provided important new insights into the regulation of cell division, which may ultimately lead to a better understanding of cancer progression. [23]
Category: Artificial Intelligence

[34] viXra:1903.0310 [pdf] submitted on 2019-03-17 01:43:24

A Right Representation of the Completeness Theorem of Propositional Logic —Towards a Truly Thinking Machine—

Authors: Atsushi Shimotani
Comments: 1 Page.

The calculation of the truth function of a propositional formula is a kind of proof.
Category: Artificial Intelligence

[33] viXra:1903.0298 [pdf] submitted on 2019-03-15 12:07:10

Adaptive Machine Learning

Authors: George Rajna
Comments: 88 Pages.

Over the past few decades, many studies conducted in the field of learning science have reported that scaffolding plays an important role in human learning. [47] The researchers trained generative competitive networks to predict the behavior of charged elementary particles. The results showed that physical phenomena can be described using neural networks highly accurately. [46] New data from the STAR experiment at the Relativistic Heavy Ion Collider (RHIC) add detail-and complexity-to an intriguing puzzle that scientists have been seeking to solve: how the building blocks that make up a proton contribute to its spin. [45] Approximately one year ago, a spectacular dive into Saturn ended NASA's Cassini mission-and with it a unique, 13-year research expedition to the Saturnian system. [44] Scientists from the Niels Bohr Institute, University of Copenhagen, and their colleagues from the international ALICE collaboration recently collided xenon nuclei, in order to gain new insights into the properties of the Quark-Gluon Plasma (the QGP)-the matter that the universe consisted of up to a microsecond after the Big Bang. [43] The energy transfer processes that occur in this collisionless space plasma are believed to be based on wave-particle interactions such as particle acceleration by plasma waves and spontaneous wave generation, which enable energy and momentum transfer. [42] Plasma particle accelerators more powerful than existing machines could help probe some of the outstanding mysteries of our universe, as well as make leaps forward in cancer treatment and security scanning-all in a package that's around a thousandth of the size of current accelerators. [41] The Department of Energy's SLAC National Accelerator Laboratory has started to assemble a new facility for revolutionary accelerator technologies that could make future accelerators 100 to 1,000 times smaller and boost their capabilities. [40]
Category: Artificial Intelligence

[32] viXra:1903.0297 [pdf] submitted on 2019-03-15 12:37:50

AI for the Study of Sites

Authors: George Rajna
Comments: 89 Pages.

This method can be used as a starting point in Taphonomy when analyzing remains in sites whose preservation does not allow distinguishing who accumulated the assemblages through the analysis of the cut or tooth marks left on the surface of the bones. [48] Over the past few decades, many studies conducted in the field of learning science have reported that scaffolding plays an important role in human learning. [47] The researchers trained generative competitive networks to predict the behavior of charged elementary particles. The results showed that physical phenomena can be described using neural networks highly accurately. [46] New data from the STAR experiment at the Relativistic Heavy Ion Collider (RHIC) add detail-and complexity-to an intriguing puzzle that scientists have been seeking to solve: how the building blocks that make up a proton contribute to its spin. [45] Approximately one year ago, a spectacular dive into Saturn ended NASA's Cassini mission-and with it a unique, 13-year research expedition to the Saturnian system. [44] Scientists from the Niels Bohr Institute, University of Copenhagen, and their colleagues from the international ALICE collaboration recently collided xenon nuclei, in order to gain new insights into the properties of the Quark-Gluon Plasma (the QGP)-the matter that the universe consisted of up to a microsecond after the Big Bang. [43] The energy transfer processes that occur in this collisionless space plasma are believed to be based on wave-particle interactions such as particle acceleration by plasma waves and spontaneous wave generation, which enable energy and momentum transfer. [42] Plasma particle accelerators more powerful than existing machines could help probe some of the outstanding mysteries of our universe, as well as make leaps forward in cancer treatment and security scanning-all in a package that's around a thousandth of the size of current accelerators. [41]
Category: Artificial Intelligence

[31] viXra:1903.0283 [pdf] submitted on 2019-03-14 10:24:06

Machine Learning Quantum Advantage

Authors: George Rajna
Comments: 44 Pages.

We are still far off from achieving Quantum Advantage for machine learning-the point at which quantum computers surpass classical computers in their ability to perform AI algorithms. [25] Physicists in the US have used machine learning to determine the phase diagram of a system of 12 idealized quantum particles to a higher precision than ever before. [24] The research group took advantage of a system at SLAC's Stanford Synchrotron Radiation Lightsource (SSRL) that combines machine learning-a form of artificial intelligence where computer algorithms glean knowledge from enormous amounts of data-with experiments that quickly make and screen hundreds of sample materials at a time. [23] Researchers at the UCLA Samueli School of Engineering have demonstrated that deep learning, a powerful form of artificial intelligence, can discern and enhance microscopic details in photos taken by smartphones. [22] Such are the big questions behind one of the new projects underway at the MIT-IBM Watson AI Laboratory, a collaboration for research on the frontiers of artificial intelligence. [21]
Category: Artificial Intelligence

[30] viXra:1903.0279 [pdf] submitted on 2019-03-14 12:31:27

Учение о системах и структурах организаций

Authors: Попов Борис Михайлович
Comments: 86 Pages. «Концерн «СОЗВЕЗДИЕ». – Воронеж, 2009. – 86 с. ISВN 978-5-900777-19-1

В монографии излагается учение об организациях, системах и структурах, базирующееся на представлении об их контекстной зависимости, как теоретических понятий, так и взаимообусловленности их существования в коммуникативном мире, как целостной триады. В рамках используемой автором теоретической платформы, − тринитарной парадигмы, несколько иное видение, по отношению к традиционному, получают понятия «информация» и «энергия». Книга рассчитана на широкий круг читателей, интересующихся проблемами организации и самоорганизации. Требования к исходным знаниям невысоки. Необходимые для понимания содержания книги сведения по математике, физике и биологии, даются по ходу изложения, которое проиллюстрировано многочисленными примерами, облегчающими усвоение прочитанного материала. Читателю гарантируется стремительный рост знаний с продвижением от первого абзаца к последнему. Может рассматриваться преподавателями технических вузов на предмет использования в качестве дополнительного учебного пособия для студентов, осваивающих курсы дисциплин проектирования организационно сложных комплексов. В т. ч. проектирования, с использованием нанотехнологий.
Category: Artificial Intelligence

[29] viXra:1903.0275 [pdf] submitted on 2019-03-14 18:28:01

Artificial Intelligent Vehicle Speed Control System Using RF Technology

Authors: Saurabh S. Kore, Kanchan Y. Hanwate, Ashwin S. Moon, Hemant P. Chavan
Comments: 4 Pages.

Traffic congestion nowadays leads to a strong degradation of the network infrastructure and the high number of the vehicle which is caused by the population. With an increase in the number of vehicles and transportation demand, traffic congestion occurs. Conventional techniques are unable to handle variable flow with time. Hence the accident ratio is increases We all know that in order to control traffic and for the safety of the public, traffic signals are used. These signals are generally seen where two or more roads are connected to each other. These traffic signal rules are not generally followed by the people either due to lack of attention or due to speeding of vehicles which causes accidents frequently. Though accidents due to lack of attention to traffic signal cannot be avoided but accidents due to speeding of vehicles near traffic signals could be avoided. In order to avoid such kind of accidents near traffic signals; ‘smart signal system’ should be introduced. This system takes the total control of the vehicle in its own band. This project consists of two separate modules, i.e. transmitter and receiver. The transmitter will be attached to signals and receiver will be attached to vehicles. The information in the form of signals are sent by the system and they are received by the receiver attached to the vehicle. The vehicle control unit automatically alerts the driver and reduces the speed automatically. This system takes total control of vehicles for a few seconds. The proposed technique uses a small system to avoid accidents and hence safety is achieved.
Category: Artificial Intelligence

[28] viXra:1903.0272 [pdf] submitted on 2019-03-15 03:43:08

Organic Network Control Systems Challenges in Building a Generic Solution for Network Protocol Optimisation

Authors: Matthias Bloch
Comments: 7 Pages.

In the last years many approaches for dynamic protocol adaption in networks have been made and proposed. Most of them deal with a particular environment, but a much more desired approach would be to design a generic solution for this problem. Creating an independent system regarding the network type it operates in and therefore the protocol type that needs to be adapted is a big issue. In this paper we want to discuss certain problems that come with this task and why they have to be taken into account when it comes to designing such a generic system. At first we will see a generic architecture approach for such a system followed by a comparison of currently existing Organic Network Control Systems for adapting protocols in a Mobile Ad-hoc network and a Peer-to-Peer network. After identifying major problems we will summarize and evaluate the achieved results.
Category: Artificial Intelligence

[27] viXra:1903.0268 [pdf] submitted on 2019-03-15 06:05:04

Some Models in the Context of Cryo-em Image Processing Applications – a Simple & Novel Suggestion Using [biips/dlib Ml/quantum Device] – as Next Generation Intelligent Electron Microscopy Informatics Platform.

Authors: Nirmal Tej Kumar
Comments: 2 Pages. Short Communication & Technical Notes

Some Models in the context of cryo-EM Image Processing Applications – A Simple & Novel Suggestion Using - [BIIPS/dlib ML/quantum Device] – as Next Generation Intelligent Electron Microscopy Informatics Platform.
Category: Artificial Intelligence

[26] viXra:1903.0265 [pdf] submitted on 2019-03-15 07:17:30

AI Predict Elementary Particle Signals

Authors: George Rajna
Comments: 87 Pages.

The researchers trained generative competitive networks to predict the behavior of charged elementary particles. The results showed that physical phenomena can be described using neural networks highly accurately. [46] New data from the STAR experiment at the Relativistic Heavy Ion Collider (RHIC) add detail-and complexity-to an intriguing puzzle that scientists have been seeking to solve: how the building blocks that make up a proton contribute to its spin. [45] Approximately one year ago, a spectacular dive into Saturn ended NASA's Cassini mission-and with it a unique, 13-year research expedition to the Saturnian system. [44] Scientists from the Niels Bohr Institute, University of Copenhagen, and their colleagues from the international ALICE collaboration recently collided xenon nuclei, in order to gain new insights into the properties of the Quark-Gluon Plasma (the QGP)-the matter that the universe consisted of up to a microsecond after the Big Bang. [43] The energy transfer processes that occur in this collisionless space plasma are believed to be based on wave-particle interactions such as particle acceleration by plasma waves and spontaneous wave generation, which enable energy and momentum transfer. [42] Plasma particle accelerators more powerful than existing machines could help probe some of the outstanding mysteries of our universe, as well as make leaps forward in cancer treatment and security scanning-all in a package that's around a thousandth of the size of current accelerators. [41] The Department of Energy's SLAC National Accelerator Laboratory has started to assemble a new facility for revolutionary accelerator technologies that could make future accelerators 100 to 1,000 times smaller and boost their capabilities. [40] The authors designed a mechanism based on the deployment of a transport barrier to confine the particles and prevent them from moving from one region of the accelerator to another.
Category: Artificial Intelligence

[25] viXra:1903.0262 [pdf] submitted on 2019-03-13 09:19:44

AI Solve Quantum Mysteries

Authors: George Rajna
Comments: 47 Pages.

Under the direction of Mobileye founder Amnon Shashua, a research group at Hebrew University of Jerusalem's School of Engineering and Computer Science has proven that artificial intelligence (AI) can help us understand the world on an infinitesimally small scale called quantum physics phenomena. [27] Researchers from the Moscow Institute of Physics and Technology teamed up with colleagues from the U.S. and Switzerland and returned the state of a quantum computer a fraction of a second into the past. [26]
Category: Artificial Intelligence

[24] viXra:1903.0260 [pdf] submitted on 2019-03-13 10:36:13

Current Trends in Extended Classifier System

Authors: Zeshan Murtza
Comments: 5 Pages.

Learning is a way which improves our ability to solve problems related to the environment surrounding us. Extended Classifier System (XCS) is a learning classifier system that use reinforcement learning mechanism to solve complex problems with robust performance. It is an accuracy-based system that works by observing environment, taking input from it and applying suitable actions. Every action of XCS gets a feedback in return from the environment which is used to improve its performance. It also has ability to apply genetic algorithm (GA) on existing classifiers and create new ones by taking cross-over and mutation which have better performance. XCS handles single step and multi-step problems by using different methods like Q-learning mechanism. The ultimate challenge of XCS is to design an implementation which arrange multiple components in a unique way to produce compact and comprehensive solution in a least amount of time. Real time implementation requires flexibility for modifications and uniqueness to cover all aspects. XCS has recently been modified for real input values and a memory management system is also introduced which enhance its ability in different kind of applications like data mining, control stock exchange. In this article, there will be a brief discussion about the parameter and components of XCS. Main part of this article will cover the extended versions of XCS with further improvements and focus on applications, usage in real environment and relationship with organic computing
Category: Artificial Intelligence

[23] viXra:1903.0236 [pdf] submitted on 2019-03-12 15:13:02

Resolving Limits of Organic Systems in Large Scale Environments: Evaluate Benefits of Holonic Systems Over Classical Approaches

Authors: Claudio Schmidt
Comments: 5 Pages.

With the rapidly increasing number of devices and application components interacting with each other within larger complex systems, classical system hierarchies increasingly hit their limit when it comes to highly scalable and possibly fluctual organic systems. The holonic approach for self-* systems states to solve some of these problems. In this paper, limits of different state-of-the-art technologies and possible solutions to those will be identified and ranked for scalability, privacy, reliability and performance under fluctuating conditions. Subsequently, the idea and structure of holonic systems will be outlined, and how to utilize the previously described solutions combined in a holonic environment to resolve those limits. Furthermore, they will be classified in the context of current multi-agent-systems (MAS). The focus of this work is located in the area of smart energy grids and similar structures, however an outlook sketches a few further application scenarios for holonic structures.
Category: Artificial Intelligence

[22] viXra:1903.0223 [pdf] submitted on 2019-03-11 09:03:33

Comparing Anytime Learning to Organic Computing

Authors: Thomas Dangl
Comments: 5 Pages.

In environments where finding the best solution to a given problem is computationally infeasible or undesirable due to other restrictions, the approach of anytime learning has become the de facto standard. Anytime learning allows intelligent systems to adapt and remain operational in a constantly changing environment. Based on observation of the environment, the underlying simulation model is changed to fit the task and the learning process begins anew. This process is expected to never terminate, therefore continually improving the set of available strategies. Optimal management of uncertainty in tasks, which require a solution in real time, can be achieved by assuming faulty yet improving output. Properties of such a system are not unlike those present in organic systems. This article aims to give an introduction to anytime learning in general as well as to show the similarities to organic computing in regards to the methods and strategies used in both domains.
Category: Artificial Intelligence

[21] viXra:1903.0215 [pdf] submitted on 2019-03-11 13:40:21

Performance Measurement of Multi-Agent Systems

Authors: Muhammad Mohiuddin
Comments: 7 Pages.

A multi-agent system can greatly increase performance and reliability of a system due to several reasons like distributed nature, responsiveness to environment, and the ability for reuse. These characteristics are associated with multi-agent systems due to their flexible and intelligent nature. All these capabilities do not come without challenges. One of the biggest challenges that arises due to the dynamic behavior of multi-agent systems is the difficulty to quantify their reliability and dependability, or in other words performance. This article discusses agents and multi-agent systems, their classification according design parameters, multiple methods of performance quantization, and factors which affect them.
Category: Artificial Intelligence

[20] viXra:1903.0202 [pdf] submitted on 2019-03-12 05:57:39

The Halting Problem of Turing is Nonsense —Towards a Truly Thinking Machine—

Authors: Atsushi Shimotani
Comments: 1 Page.

The halting problem of Turing is nonsense. The program will halt if the input is the self. It is not contradiction.
Category: Artificial Intelligence

[19] viXra:1903.0189 [pdf] submitted on 2019-03-10 16:07:44

Using Self-Awareness in Decentralized Computing Systems

Authors: Florian Maier
Comments: 4 Pages

The term self-awareness in technological systems has been discussed many years now. There is no commonly agreed definition of the term self-aware in biological or psychological meaning therefore there are many different definitions all stating different aspects and levels of what we call self-aware in the biological world. In addition the system should be characterized by properties such as: robustness, decentralization, flexibility and self-adapting. In the past this was often achieved by designing good and robust but also complex algorithms which often lead to unnecessary overhead and hard to fix runtime bugs. Using self-aware components can turn such algorithmic systems into organic computing systems, offering better scalability, more robustness of the global state and less unnecessary overhead in the communication between different components in the decentralized system. On the counterpart such a system might work in a way that cannot be fully understood by humans in a reasonable time leading to other problems such as trust issues in the system or unwanted behavior in the global state of the system. The goal of this article is to state out how decentralized computing systems can benefit from self-aware approaches.
Category: Artificial Intelligence

[18] viXra:1903.0186 [pdf] submitted on 2019-03-10 19:47:08

Advancements of Deep Q-Networks

Authors: Bastian Birkeneder
Comments: 5 Pages.

Deep Q-Networks first introduced a combination of Reinforcement Learning and Deep Neural Networks at a large scale. These Networks are capable of learning their interactions within an environment in a self-sufficient manor for a wide range of applications. Over the following years, several extensions and improvements have been developed for Deep Q-Networks. In the following paper, we present the most notable developments for Deep Q-Networks, since the initial proposed algorithm in 2013.
Category: Artificial Intelligence

[17] viXra:1903.0177 [pdf] submitted on 2019-03-11 04:35:04

Generalized Deng Entropy

Authors: Fan Liu, Xiaozhuan Gao, Yong Deng
Comments: 14 Pages.

Dempster-Shafer evidence theory as an extension of Probability has wideapplications in many fields. Recently, A new entropy called Deng entropywas proposed in evidence theory. Deng Entropy as an uncertain measurein evidence theory. Recently, some scholars have pointed out that DengEntropy does not satisfy the additivity in uncertain measurements. However,this irreducibility can have a huge effect. In more complex systems, thederived entropy is often unusable. Inspired by this, a generalized entropy isproposed, and the entropy implies the relationship between Deng entropy,R ́enyi entropy, Tsallis entropy.
Category: Artificial Intelligence

[16] viXra:1903.0168 [pdf] submitted on 2019-03-09 09:58:15

Organic Traffic Control with Dynamic Route Guidance as a Measure to Reduce Exhaust Emissions in Comparison Organic Traffic Control Mit Dynamic Route Guidance Als Maßnahme Zur Reduzierung Von Abgasemissionen im Vergleich

Authors: Christian Frank
Comments: 5 pages, 2 figures, language: German

In this paper an Organic Traffic Control system with Dynamic Route Guidance functionality is being looked at regarding its emission-reducing effect on road traffic. This system will be compared to other environmental measures, namely Low Emission Zones, driving bans and hardware upgrades, with respect to its effect on emissions and other criteria. Results from existing literature and a few calculations are used for this comparison. The sparse data allows for only a few quantitive comparisons. Qualitative comparisons show that this system has the potential to effectively lower emission in its area of effect. It reduces the quantity of all exhaust gases and additionally fuel consumption, without disadvantages for certain road users. This is not the case with the comparative measures. ----- In dieser Arbeit wird ein Organic Traffic Control System mit Dynamic Route Guidance Funktionalität hinsichtlich seiner emissionsreduzierenden Wirkung im Verkehr betrachtet. Dieses System wird mit anderen Umweltmaßnahmen, namentlich Umweltzonen, Fahrverboten und Hardwarenachrüstungen, hinsichtlich Wirkung und weiterer Kriterien verglichen. Es werden hierzu Daten und Ergebnisse aus der bestehenden Literatur verwendet und einige wenige Rechnungen durchgeführt. Die Datenlage erlaubt nur teilweise quantitative Vergleiche. Qualitativ zeigt sich, dass das System Potential bietet, effektiv innerhalb seines Installationsbereichs Emissionen zu senken. Es reduziert die Menge aller Abgase und zusätzlich den Spritverbrauch, ohne dass dabei Nachteile für bestimme Verkehrsteilnehmer entstehen. Dies ist bei den Vergleichsmaßnahmen jeweils nicht der Fall.
Category: Artificial Intelligence

[15] viXra:1903.0155 [pdf] submitted on 2019-03-10 05:27:02

The Principle, Communication Efficiency and Privacy Issues of Federated Learning

Authors: Hakan Uzuner
Comments: 5 Pages.

Standard machine learning approaches require a huge amount of training data to be stored centralized in order to feed the learning algorithms. Keeping and using data centralized brings many negative aspects with it. Those aspects can be inefficient communication between the centralized data center and the clients producing the data, privacy issues and quick usability of the profits and results of the training. Google’s new approach, federated learning, on the other hand tackles all these problems. The training data is kept decentralized at the client’s devices while communicating only with small updates of the common model. This method allows for optimizations of communication, keeping the privacy of users involved in the process and providing quick usability of the model’s process. In this paper I will explain how the federated learning principle works. Further on, I will give a small insight on optimization possibilities of communication efficiency as well as on privacy issues involved in machine learning processes and how those can be solved using federated learning principles. Additionally, I will show the connection between the federated learning concept and organic computing.
Category: Artificial Intelligence

[14] viXra:1903.0139 [pdf] submitted on 2019-03-09 00:40:07

CoqTP-OCaml-Java Based Some Important Applications – A Simple Suggestion & an Insight as Short Communication.

Authors: Nirmal Tej Kumar
Comments: 2 Pages. Short Communication & Technical Notes

CoqTP-OCaml-Java Based Some Important Applications – A Simple Suggestion & an Insight as Short Communication.
Category: Artificial Intelligence

[13] viXra:1903.0138 [pdf] submitted on 2019-03-09 01:41:05

A Survey on Reinforcement Learning for Dialogue Systems

Authors: Isabella Graßl
Comments: 6 Pages.

Dialogue systems are computer systems which com- municate with humans using natural language. The goal is not just to imitate human communication but to learn from these interactions and improve the system’s behaviour over time. Therefore, different machine learning approaches can be implemented with Reinforcement Learning being one of the most promising techniques to generate a contextually and semantically appropriate response. This paper outlines the current state-of- the-art methods and algorithms for integration of Reinforcement Learning techniques into dialogue systems.
Category: Artificial Intelligence

[12] viXra:1903.0135 [pdf] submitted on 2019-03-09 05:32:36

A Survey on Classification of Concept Drift with Stream Data

Authors: Shweta Vinayak Kadam
Comments: 7 Pages.

Usually concept drift occurs in many applications of machine learning. Detecting a concept drift is the main challenge in a data stream because of the high speed and their large size sets which are not able to fit in main memory. Here we take a small look at types of changes in concept drift. This paper discusses about methods for detecting concept drift and focuses on the problems with existing approaches by adding STAGGER, FLORA family, Decision tree methods, meta-learning methods and CD algorithms. Furthermore, classifier ensembles for change detection are discussed.
Category: Artificial Intelligence

[11] viXra:1903.0133 [pdf] submitted on 2019-03-07 07:07:19

Deep Learning Holography

Authors: George Rajna
Comments: 48 Pages.

Digital holographic microscopy is an imaging modality that can digitally reconstruct the images of 3-D samples from a single hologram by digitally refocusing it through the entire 3-D sample volume. [27] Deep learning, which uses multi-layered artificial neural networks, is a form of machine learning that has demonstrated significant advances in many fields, including natural language processing, image/video labeling and captioning. [26]
Category: Artificial Intelligence

[10] viXra:1903.0128 [pdf] submitted on 2019-03-07 09:45:47

XCS-O/C: Das Extended Classifier System XCS in einer Observer/Controller Architektur

Authors: Alexander Paßberger
Comments: 7 Pages. German

Organic Computing stellt eine moderne Variante des Entwerfen autonomer Systeme dar, bei der Entscheidungen zur Laufzeit vom System selbst getroffen werden. Das Steuern der ausgeführten Aktionen benötigt in einem solchen System Möglichkeiten zur Selbstverbesserung. Ein möglicher Ansatz zur Problemlösung sind Learning Classifier Systeme, speziell das Extended Classifier System XCS. Die Arbeit liefert eine detaillierte Beschreibung der Abläufe im Extended Classifier System XCS und der nötigen Änderungen zum Integrieren in ein autonomes organisches System. Das Learning Classifier System wird hierzu in eine generische Observer/Controller-Architektur eingebettet, eine der fundamentalen Designarchitekturen im Bereich Organic Computing.
Category: Artificial Intelligence

[9] viXra:1903.0127 [pdf] submitted on 2019-03-07 10:01:05

AI Ask for Human Help

Authors: George Rajna
Comments: 63 Pages.

This is a demonstration of a situation where an AI algorithm working together with a human can reap the benefits and efficiency of the AI's good decisions, without being locked into its bad ones. [33] The deep learning analysis has revealed that the extinct hominid is probably a descendant of the Neanderthal and Denisovan populations. [32] Our team at IBM Research – India collaborated with the IBM MetroPulse team to bring such first-of-a-kind, AI-driven capabilities to MetroPulse, an industry platform that brings together voluminous market, external and client datasets. [31]
Category: Artificial Intelligence

[8] viXra:1903.0121 [pdf] submitted on 2019-03-08 02:01:34

Online Transfer Learning and Organic Computing for Deep Space Research and Astronomy

Authors: Sadanandan Natarajan
Comments: 6 Pages.

Deep space exploration is the pillars within the field of outer space analysis and physical science. The amount of knowledge from numerous space vehicle and satellites orbiting the world of study are increasing day by day. This information collected from numerous experiences of the advanced space missions is huge. These information helps us to enhance current space knowledge and the experiences can be converted and transformed into segregated knowledge which helps us to explore and understand the realms of the deep space.. Online Transfer Learning (OTL) is a machine learning concept in which the knowledge gets transferred between the source domain and target domain in real time, in order to help train a classifier of the target domain. Online transfer learning can be an efficient method for transferring experiences and data gained from the space analysis data to a new learning task and can also routinely update the knowledge as the task evolves.
Category: Artificial Intelligence

[7] viXra:1903.0120 [pdf] submitted on 2019-03-08 02:36:18

A Discussion of Detection of Mutual Influences Between Socialbots in Online (Social) Networks

Authors: Stefanie Urchs
Comments: 6 Pages.

Many people organise themselves online in social networks or share knowledge in open encyclopaedias. However, these networks do not only belong to humans. A huge variety of socialbots that imitate humans inhabit these and are connected to each other. The connections between socialbots lead to mutual influences between them. If the influence socialbots have on each other are too big they adapt the behaviour of the other socialbot and get worse in imitating humans. Therefore, it is necessary to detect when socialbots are mutually influencing each other. For a better overview socialbots in the social networks Facebook, Twitter and in the open encyclopaedia Wikipedia are observed and the mutual influences between them detected. Furthermore, this paper discusses how socialbots could handle the detected influences.
Category: Artificial Intelligence

[6] viXra:1903.0117 [pdf] submitted on 2019-03-08 04:47:40

A Survey on Different Mechanisms to Classify Agent Behavior in a Trust Based Organic Computing Systems

Authors: Shabhrish Reddy Uddehal
Comments: 9 Pages.

Organic Computing (OC) systems vary from traditional software systems, as these systems are composed of a large number of highly interconnected and distributed subsystems. In systems like this, it is not possible to predict all possible system configurations and to plan an adequate system behavior entirely at design time. An open/decentralized desktop grid is one example, Trust mechanisms are applied on agents that show the following Self-X properties (Self-organization, Self-healing, Self-organization and so on). In this article, some mechanisms that could help in the classification of agents behavior at run time in trust-based organic computing systems are illustrated. In doing so, isolation of agents that reduce the overall systems performance is possible. Trust concept can be used on agents and then the agents will know if their interacting agents belong to the same trust community and how trustworthy are they. Trust is a significant concern in large-scale open distributed systems. Trust lies at the core of all interactions between the agents which operate in continuously varying environments. Current research leads in the area of trust in computing systems are evaluated and addressed. This article shows mechanisms discussed can successfully identify/classify groups of systems with undesired behavior.
Category: Artificial Intelligence

[5] viXra:1903.0089 [pdf] submitted on 2019-03-05 09:07:04

Deep Meta-Learning and Dynamic Runtime Exploitation of Knowledge Sources for Traffic Control

Authors: Sandra Ottl
Comments: 7 Pages.

In the field of machine learning and artificial intelligence, meta-learning describes how previous learning experiences can be used to increase the performance on a new task. For this purpose, it can be investigated how prior (similar) tasks have been approached and improved, and knowledge can be obtained about achieving the same goal for the new task. This paper outlines the basic meta-learning process which consists of learning meta-models from meta-data of tasks, algorithms and how these algorithms perform on the respective tasks. Further, a focus is set on how this approach can be applied and is already used in the context of deep learning. Here, meta-learning is concerned with the respective machine learning models themselves, for example how their parameters are initialised or adapted during training. Also, meta-learning is assessed from the viewpoint of Organic Computing (OC) where finding effective learning techniques that are able to handle sparse and unseen data is of importance. An alternative perspective on meta-learning coming from this domain that focuses on how an OC system can improve its behaviour with the help of external knowledge sources, is highlighted. To bridge the gap between those two perspectives, a model is proposed that integrates a deep, meta-learned traffic flow predictor into an organic traffic control (OTC) system that dynamically exploits knowledge sources during runtime.
Category: Artificial Intelligence

[4] viXra:1903.0086 [pdf] submitted on 2019-03-05 15:43:42

Novelty Detection Algorithms and Their Application in Industry 4.0

Authors: Christoph Stemp
Comments: 7 Pages.

Novelty detection is a very important part of Intelligent Systems. Its task is to classify the data produced by the system and identify any new or unknown pattern that were not present during the training of the model. Different algorithms have been proposed over the years using a wide variety of different technologies like probabilistic models and neural networks. Novelty detection and reaction is used to enable self*-properties in technical systems to cope with increasingly complex processes. Using the notion of Organic Computing, industrial factories are getting more and more advanced and intelligent. Machines gain the capability of self-organization, self-configuration and self-adaptation to react to outside influences. This survey paper looks at the state-of-the-art technologies used in Industry 4.0 and assesses different novelty detection algorithms and their usage in such systems. Therefore, different data-sources and consequently applications for potential novelty detection are analyzed. Three different novelty detection algorithms are then present using different underlying technologies and the applicability of these algorithms in combination with the defined scenarios is analyzed.
Category: Artificial Intelligence

[3] viXra:1903.0045 [pdf] replaced on 2019-03-17 15:32:33

How Deep Is Deep Enough? HDDE9-2S A CNN Network With 12 Layers Reached Higher Accuracy In Image Classification

Authors: Guangqun Chen
Comments: 10 Pages.

In this paper I studied SfNet further. I reduced layers of SfNet, created a group of networks: HDDE6 series and HDDE9 series. HDDE6-3 and HDDE6-2S have 9 layers, HDDE9-2 has 11 layers, HDDE9-2S and HDDE9-3 have 12 layers. In my experiments on dataset CALTECH-256, compared to SFNet, the classification accuracy of HDDE9-3 increases about 4.35%, the classification accuracy of HDDE9-2 increases about 3.07%, the classification accuracy of HDDE9-2S increases about 5.55%. Compared to VGG-16, the classification accuracy of HDDE9-3 increases from 83.63% to 90.73%, the classification accuracy of HDDE9-2 increases to 89.81%, the classification accuracy of HDDE9-2S increases to 92.28%. For HDDE6 series, features extraction part uses only 1 convolution layer and 5 MixedSCLayers, has much less parameters, running speed is much faster, the accuracy of HDDE6-3 is the same as HDDE9-3's on CALTECH-256. All the improvements are due to the use of Structure Composing Layers.
Category: Artificial Intelligence

[2] viXra:1903.0012 [pdf] submitted on 2019-03-02 04:32:21

A Survey for Testing Self-organizing, Adaptive Systems in Industry 4.0

Authors: Caterina Rotondo
Comments: 6 Pages.

Complexity in technical development increases rapidly. Regular system are no longer able to fulfill all the requirements. Organic computing systems are inspired by how complexity is mastered in nature. This leads to a fundamental change in software engineering for complex systems. Based on machine learning techniques, a system develops self*-properties which allows it to make decisions at runtime and to operate with nearly no human interaction. Testing is a part of the software engineering process to ensure the functionality and the quality of a system. But when using self-organizing, adaptive systems traditional testing approaches reach their limits. Therefore, new methods for testing such systems have to be developed. There exist already a lot of different testing approaches. Most of them developed within a research group. Nevertheless, there is still a need for further discussion and action on this topic. In this paper the challenges for testing self-organizing, adaptive systems are specified. Three different testing approaches are reviewed in detail. Due to the ongoing fourth industrial revolution it is discussed which of these approaches would fit best for testing industrial manufacturing robots.
Category: Artificial Intelligence

[1] viXra:1903.0006 [pdf] submitted on 2019-03-01 03:32:14

Multi-Agent Reinforcement Learning - From Game Theory to Organic Computing

Authors: Maurice Gerczuk
Comments: 6 Pages.

Complex systems consisting of multiple agents that interact both with each other as well as their environment can often be found in both nature and technical applications. This paper gives an overview of important Multi-Agent Reinforcement Learning (MARL) concepts, challenges and current research directions. It shortly introduces traditional reinforcement learning and then shows how MARL problems can be modelled as stochastic games. Here, the type of problem and the system configuration can lead to different algorithms and training goals. Key challenges such as the curse of dimensionality, choosing the right learning goal and the coordination problem are outlined. Especially, aspects of MARL that have previously been considered from a critical point of view are discussed with regards to if and how the current research has addressed these criticism or shifted their focus. The wide range of possible MARL applications is hinted at by examples from recent research. Further, MARL is assessed from an Organic Computing point of view where it takes a central role in the context of self-learning and self-adapting systems.
Category: Artificial Intelligence