Lernergebnisse
Engineering disciplines now widely use machine learning and deep learning for system monitoring, fault detection, data-driven decision support, and harnessing big data opportunities. This module teaches advanced deep learning concepts and their Python implementation using standard libraries. Real-world engineering examples are employed to emphasize comprehension of crucial concepts in feed-forward, convolutional, and recurrent deep neural networks, including sequence classification, image classification, and object recognition.
Upon successful completion of the module, students will acquire the following:
Knowledge:
- Advanced understanding of (un-)supervised deep learning methods, including their structure and functionality.
- Familiarity with error backpropagation, various optimization algorithms, and their unique characteristics.
- Proficiency in architectural design and conception of deep learning methods.
- Knowledge of essential neural training parameters, regularization techniques, and training strategies.
Skills:
- Statistical characterization and evaluation of large, high-dimensional datasets.
- Handling unstructured data using convolutional and recurrent neural networks.
- Effective visualization of large, high-dimensional datasets.
- Implementation of core operations and key neural architectures from scratch.
- Utilization of popular programming libraries in Python.
Competencies:
- Exploratory analysis of extensive unstructured datasets.
- Feature engineering for sequential data and transformation into structured formats.
- Selection of appropriate deep learning neural architectures for structured and unstructured data.
- Evaluation of predictions, assessing bias and variance in complex deep neural networks.
- Assessment of risks, environmental impact, and technological implications.
The course teaches 60% knowledge & understanding, 20% analysis & methodology, and 20% programming.
