Advances in Real-Time and Autonomous Systems : Proceedings of the 15th International Conference on Autonomous Systems / edited by Herwig Unger, Marcel Schaible.

Format:

Book

Author/Creator:

Unger, Herwig.

Contributor:

Schaible, Marcel.

Series:

Lecture Notes in Networks and Systems, 2367-3389 ; 1009

Language:

English

Subjects (All):

Engineering--Data processing.

Engineering.

Application software.

Computational intelligence.

Data Engineering.

Computer and Information Systems Applications.

Computational Intelligence.

Local Subjects:

Data Engineering.

Computer and Information Systems Applications.

Computational Intelligence.

Physical Description:

1 online resource (185 pages)

Edition:

1st ed. 2024.

Place of Publication:

Cham : Springer Nature Switzerland : Imprint: Springer, 2024.

Summary:

The 15th International conference on "Autonomous Systems" is addressing an important topic for computing systems and communication networks due to increasingly complex environments they are working in. Therefore, this conferences addresses methods and solutions enabling such systems to independently adapt themselves to changed environmental conditions, are able to learn from different sources and to process external requests dependably, safely and timely. The solutions presented here range from hardware to system design to individual applications. The book contains the results of the researchers presented at this conference, which is supported by GI (German Society of informatics) and other organisations Real-Time Systems Expert Committees. The target audience is students and researchers in computer science and automation technology, engineers, programmers and users of automation and communication systems.

Contents:

Intro

Preface

Organization

Contents

Algorithmic Foundations of Reinforcement Learning

1 Introduction

2 Reinforcement Learning Fundamentals

2.1 Markov Decision Processes (MDPs)

2.2 Solving an MDP: Dynamic Programming

2.3 Model-Free Prediction

3 Reinforcement Learning Algorithms

3.1 Model-Free Reinforcement Learning

3.2 Deep Q-Learning

3.3 Policy Methods

3.4 Actor-Critic Methods

3.5 Off-Policy Deep RL and Continuous Control

3.6 Conclusion and Summary

References

Autonomous Emergency Landing of an Aircraft in Case of Total Engine-Out

2 High-Key/Low-Key Heuristics

3 Computer-Based Avionics Systems

4 Optimized Emergency Approach Routes that Provide Altitude Division

4.1 General Problem Analysis

4.2 Compensation of Wind Influence

4.3 Database of Emergency Landing Fields

4.4 Safe2Land App (Version 1)

4.5 Safe2Land App (Version 2)

5 Conclusion

The Safety-Related Real-Time Language SafePEARL

2 Major Language Properties

2.1 Program Structure

2.2 Algorithmics

2.3 Process Data Input and Output

2.4 Real-Time Programming

3 PEARL for Distributed Systems

4 Safety-Related PEARL

4.1 Verification-Oriented Language Subsets

4.2 SIL1: Constructive Preclusion of Error Sources

4.3 SIL2: Predictable Time Behaviour

4.4 SIL3: Function Block Diagrams

4.5 SIL4: Cause-Effect Tables

4.6 Synopsis of the Language Subsets

5 Safe Sequential Function Charts

6 PEARL as Specification Language

Meta-Learning for Time Series Analysis and/or Forecasting: Concept Review and Comprehensive Critical Comparative Survey

1.1 Background and Motivation

1.2 Problem Statement and Research Questions

2 Fundamentals of Meta-Learning

2.1 Fundamentals Concepts of Meta-Learning.

2.2 Meta-Learning for Time Series Data: Classification Problems vs. Regression Problems

2.3 Discussion of the Different Meta-Learning Taxonomies and Approaches

2.4 Critical Discussion of the Meta-Learning Taxonomies and Approaches with Focusing on Time Series Data

3 Comprehensive Review of Meta-Learning in TSA/TSF with Regard to ZSL, OSL and FSL (RQ2)

3.1 Relevance of Meta-Learning in TSA and TSF

3.2 Critical Discussion of ZSL, OSL and FSL in Meta-Learning

3.3 Meta-Learning Example Formulation of ZSL, OSL and FSL for TSA and TSF

3.4 Pipeline Discussion for ZSL, OSL and FSL for TSF

3.5 Comprehensive Review of Meta-Learning with Regard to ZSL, OSL and FSL for TSA/TSF

4 Illustrative Implementations of ZSL, OSL and FSL for TSF (RQ3)

4.1 Dataset Presentation

4.2 Setup of the Implementation

4.3 Discussion of Performance Metrics of Relevance to ZSL/OSL/FSL

4.4 ML Baseline Implementation: Standard Encoder

4.5 ZSL Forecasting Implementation: ForecastPFN

4.6 OSL Forecasting Implementation: Siamese Networks

4.7 FSL Forecasting Implementation: Reptile

4.8 Comparative Discussion of the Results

5 Challenges and Limitations of Current Meta-Learning Techniques and Directions for Future Research in the Context of TSF/TSA

5.1 Challenges and Limitations of Meta-Learning for TSA/TSF

5.2 Suggesting Potential Avenues and Directions for Future Research in TSA/TSF

6 Conclusion

Ensemble Learning with Physics-Informed Neural Networks for Harsh Time Series Analysis

2 Physics-Informed Neural Networks (PINN): Mechanics, Recent Extensions, and Potentials

2.1 Basics of Physics-Informed Neural Networks

2.2 Recent Extensions of PINN

2.3 Potentials of PINN in Solving Complex Problems.

3 Challenges in Analyzing and Forecasting Complex and Multivariate Harsh Time Series: Focusing on Traffic and Weather

3.1 Challenges in Traditional Analysis of Time Series

3.2 Challenges in Forecasting Time Series

3.3 Special Considerations for Traffic and Weather Time Series

3.4 The Need for Advanced Approaches

4 Technical Review of Advanced Physics-Informed Neural Networks for Reliable Analysis and Forecasting of Complex Harsh Time Series in Traffic and Weather

4.1 Advanced Physics-Informed Neural Networks

4.2 Classification and Anomaly Detection in Traffic Data

4.3 Forecasting in Traffic and Weather

4.4 Performance Comparison with Traditional Approaches

5 Ensemble Learning for Enhanced Performance of Physics-Informed Neural Networks in Complex Time Series Analysis and Forecasting

5.1 Ensemble Learning Techniques and PINN

5.2 Ensemble Architectures with Selected DL Models

6 Enhancing Explainability and Interpretability in Time Series Analysis and Forecasting by Physics-Informed Neural Networks

6.1 Explainability in Time Series Analysis

6.2 Interpretability in Time Series Forecastings

6.3 Case Studies: Examples of Explainable and Interpretable Outcomes

6.4 Bridging the Gap Between Data-Driven and Physics-Informed Models

6.5 Challenges and Future Directions

7 Advantages of Using PINN for Traffic and Weather Data

8 Future Work

Language Meets Vision: A Critical Survey on Cutting-Edge Prompt-Based Image Generation Models

2 Description of the Concept of Language-Driven Image Generation Models

2.1 Background and Evolution of Prompt-Based Image Generation

2.2 Motivation and Expectations of Prompt-Based Image Generation

3 Comprehensive Specification Book for Prompt-Based Generative Models

3.1 User Input (Natural Language Prompts) Guidelines.

3.2 Training Data

3.3 Model Capabilities

3.4 Output Evaluation

3.5 Training Pipeline

4 Historical Overview Review of Deep Learning in Image Generation

5 Critical Comparative Analysis of the State-of-the-Art Prompt-Based Image Generation Models

5.1 Review of the State-of-the-Art Models

5.2 Comparative Evaluation

5.3 Gap Analysis: Identifying Limitations

6 Training Techniques for Generative Models

6.1 Challenges in Training

7 Conclusion and Future Work

Blockchain and Beyond - A Survey on Scalability Issues

2 Brief History of Blockchain

2.1 Bitcoin - A (Partial) Success Story

3 Optimizations Regarding Scalability

3.1 Approaches to Mitigate the Scalability Limitations of Traditional Blockchains

3.2 Blockchain Trilemma

3.3 Layer 2 Solutions

3.4 Payment Channel

3.5 Side Chains

3.6 Optimistic Rollup

4 Directed Acyclic Graph as DLT

4.1 Non-linear Structure

4.2 On-tangle Voting

4.3 Conflict Resolution

4.4 Side Effects and Their Treatment

4.5 Bait-and-Switch Attack

Emotion-Aware Chatbots: Understanding, Reacting and Adapting to Human Emotions in Text Conversations

1.1 Chatbots

1.2 The Case for Self-adapting Emotionally Intelligent Chatbots

2 Emotion in AI

2.1 Overview

2.2 Learning Based on Emotion

3 Methodology

3.1 Search Method

3.2 Categorization of Approaches

4 Results

4.1 Adaption of Agent's Emotion Based on User's Emotion

4.2 Response Selection Based on User's Emotion

4.3 Learning Based on Emotion

Author Index.

Other Format:

Print version: Unger, Herwig Advances in Real-Time and Autonomous Systems

ISBN:

9783031614187

OCLC:

1440128910