Data Science - Analytics and Applications(2022) [] [9783658362959]

Preface
Data science & AI depend on smart ecosystems to provide society with innovative solutions
Data boost industry-academia link
Organization
Contents
PART 1 RESEARCH TRACK
	German Abstracts
	Full Papers - Peer Reviewed
	Predictive Maintenance and Hyperparameter Optimization in the Industrial Setting
	1 Evaluation of Hyperparameter-Optimization Approaches in an Industrial Federated Learning System
		I. INTRODUCTION
		II. ALGORITHMIC CHALLENGES AND FORMAL REQUIREMENTS FOR INDUSTRIAL ASSETS
			A. Problem Formulation
			B. Federated Learning
			C. Industrial Federated Learning
		III. HYPERPARAMETER OPTIMIZATION APPROACHES IN AN IFL SYSTEM
		IV. DATA, ALGORITHMS AND EXPERIMENTS
			A. Data
			B. Algorithms
			C. Experiments
		V. EXPERIMENTAL RESULTS
		REFERENCES
	2 Towards Robust and Transferable IIoT Sensor based Anomaly Classification using Artificial Intelligence
		I. INTRODUCTION
		II. RELATED WORK
			A. IIoT Sensor Systems
			B. Time Series Classification
		III. IIOT MEASUREMENT SYSTEM
			A. Industrial Asset
			B. Sensor Measurement System
			C. Machine Learning System and Deployment
			D. Model Dissemination and Requirements
		IV. PROPOSED MACHINE LEARNING PIPELINE
			A. End-to-End-Approach: ROCKET.
		V. MEASUREMENT SETUP AND DATA SET DESCRIPTION
		VI. MACHINE LEARNING EXPERIMENTS
		VII. EXPERIMENTAL RESULTS AND CONCLUSION
		VIII. CONCLUSION AND FUTURE WORK
		REFERENCES
	3 Data-driven Cut-off Frequency Optimization for Biomechanical Sensor Data Pre-Processing
		I. INTRODUCTION
		II. METHODS
			A. The FcOpt method
			B. Evaluation
		III. RESULTS
			A. FcOpt sampling rate robustness
			B. FcOpt temporal accuracy evaluation
		IV. DISCUSSION
		V. CONCLUSION
		ACKNOWLEDGMENT
		REFERENCES
	Deep Learning based Anomaly Detection and Scene Classification
	4 A Low-Complexity Deep Learning Framework For Acoustic Scene Classification
		I. INTRODUCTION
		II. THE LOW-COMPLEXITY DEEP LEARNING FRAMEWORK PROPOSED
			A. Our baseline
			B. Ensemble of multiple spectrogram inputs
			C. Model compression methods applied to the CNN-7 network
		III. EVALUATION SETTING
			A. TAU Urban Acoustic Scenes 2020 Mobile, the Development [31] and Evaluation [32] datasets (DCASE 2021 Task1A)
			B. Deep learning framework implementation
			C. Metric for evaluation
			D. Optimize the proposed framework by evaluating factors of time length and data augmentation
		IV. EXPERIMENTAL RESULTS AND DISCUSSION
			A. Performance comparison between DCASE baseline and the CNN-7 baseline with or without using model compression methods
			B. Effect of time length, data augmentation, spectrogram input
			C. Evaluate ensemble of different spectrogram inputs
			D. Compare with the state-of-the-art systems
		V. CONCLUSION
		ACKNOWLEDGEMENT
		REFERENCES
	5 Anomaly Detection in Medical Imaging - A Mini Review
		I. INTRODUCTION
		II. METHOD
		III. RESULTS
		IV. DISCUSSION
		ACKNOWLEDGMENT
		REFERENCES
	6 Deep Learning Frameworks Applied For Audio-Visual Scene Classification
		I. INTRODUCTION
		II. DEEP LEARNING FRAMEWORKS PROPOSED
			A. Audio-based deep learning frameworks
			B. Visual-based deep learning frameworks
		III. EVALUATION SETTING
			A. TAU Urban Audio-Visual Scenes 2021 dataset [13] (Development and Evaluation datasets)
			B. Deep learning framework implementation
			C. Metric for evaluation
			D. Late fusion strategy for multiple predicted probabilities
		IV. EXPERIMENTAL RESULTS AND DISCUSSION
			A. Analysis of audio-based deep learning frameworks for scene classification
			B. Analysis of visual-based deep learning frameworks for scene classification
			C. Combine both visual and audio features for scene classification
			D. Early detecting scene context
			E. Compare with the state-of-the-art systems
			V. CONCLUSION
		ACKNOWLEDGEMENT
		REFERENCES
	Security and Data Integrity in Machine Learning
	7 Toward Applying the IEC 62443 in the UAS for Secure Civil Applications
		I. INTRODUCTION
			A. Related Work
		II. APPLYING IEC 62443 SECURITY STANDARD IN UAS
			A. Assets Identification
			B. Identify Security Zones
			C. Risk Analysis
			D. Risk Evaluation and Security Target Estimation
			E. Apply Security Requirements and Map FRs with STRIDE
		III. CONCLUSION AND FUTURE WORK
		IV. ACKNOWLEDGEMENT
		REFERENCES
	8 IAIDO: A Framework for Implementing Integrity-Aware Intelligent Data Objects
		I. INTRODUCTION
		II. RELATED WORK
		III. PRELIMINARIES
		IV. EXPERIMENTAL EVALUATION
			A. User-Defined Experimental Constraints
			B. food_info Intelligent Constraints
			C. food_info Sub-Class Intelligent Constraints
			D. Experimental Results
			E. Reasoning with Quarantine
		V. CONCLUSIONS
		VI. FUTURE WORK
		REFERENCES
	Natural Language Processing based Optimization Methods
	9 Reducing Operator Overload with Context-Sensitive Event Clustering
		I. INTRODUCTION
		II. RELATED WORK
		III. METHODOLOGY
			A. EventType2Vec
			B. Agglomerative Clustering
			C. Co-Occurrence Detection
		IV. EVALUATION AND RESULTS
			A. Dataset and Preprocessing
			B. Clustering
			C. Process Mining
		VI. CONCLUSION
		ACKNOWLEDGEMENTS
		REFERENCES
	10 Dynamic Review-based Recommenders
		I. INTRODUCTION
		II. RELATED WORK
		III. DYNAMIC REVIEW-BASED RECOMMENDERS (DRR)
			A. Dynamic Model of Review Sequences
			B. Dynamic Model of Review Content
			C. Combining temporal and summary representations
			D. Rating Model
			E. DRR Loss function
		IV. CAUSALITY
		V. EXPERIMENTS AND RESULTS
		VI. CONCLUSION AND FEATURE WORK
		ACKNOWLEDGMENT
		REFERENCES
INDUSTRY TRACK
	Abstracts
	Provided Papers - Non Reviewed
	11 Beyond Desktop Computation: Challenges in Scaling a GPU Infrastructure
		I. INTRODUCTION
		II. CLOUD VS. ON-PREMISES COMPUTING
			A. Costs
			B. GDPR
			C. Other issues with Cloud Resources
		III. REQUIREMENTS TO OUR ON-PREMISES INFRASTRUCTURE
		IV. INFRASTRUCTURE ARCHITECTURE
		V. CLUSTER SETUP
		VI. FUTURE ADAPTIONS
		VII. DISCUSSION AND CONCLUSION
		REFERENCES
	12 Strategic Approaches to the Use of Data Science in SMEs
		I. INTRODUCTION
		II. DATA SCIENCE MATURITY MODEL
		III. MULTI-CASE STUDY: DATA SCIENCE IN SMES IN SALZBURG
		IV. RESULTS
			A. SMEs
			B. Universities
			C. Government
		V. CONCLUSION
		REFERENCES
	13 Minimal-Configuration Anomaly Detection for IIoT Sensors
		I. INTRODUCTION
		II. INDUSTRIAL REQUIREMENTS AND DESIGN RATIONAL
		III. DATA SET CREATION
		IV. EXPERIMENTAL SETUP
		V. PRELIMINARY RESULTS
		VI. DISCUSSION
		REFERENCES
	14 Flexible Systems to Reach High Security Levels in the Communication with Machines and in their Maintenance
		I. MOTIVATION
		II. TRUST LEVELS IN PRODUCTION ENVIRONMENTS
		III. SYSTEM ARCHITECTURE