InceptionTime identifies the optimal receptive field size for AlexNet in time series classification. It addresses the challenge of determining how much context (data points) AlexNet needs to effectively classify time series data. The project uses a modified Inception network to systematically explore different receptive field sizes, providing insights into the network's behavior and performance.
A key contribution is the systematic exploration of receptive fields, revealing optimal contexts for AlexNet in time series classification. The code is designed for reproducibility and supports experimentation on various datasets and network configurations. It provides a thorough analysis of the impact of receptive field size on training time and accuracy.
- Receptive Field Exploration: Systematically searches for the optimal receptive field size for time series data classification.
- AlexNet Implementation: Provides a modified Inception network for time series classification tasks.
- Dataset Support: Compatible with the UCR Time Series Classification Archive and other time series datasets.
The project is complete and well-documented, with a clear README and usage instructions. It includes results for the UCR archive and provides code for generating various diagnostic plots (training time, receptive field). Recent commits indicate ongoing maintenance and updates.
This project benefits researchers and practitioners working with time series classification. It offers insights into the importance of receptive field size for deep learning models and provides a readily usable implementation of InceptionTime. It enables users to evaluate the performance of AlexNet with different receptive field sizes and understand its impact on training efficiency and classification accuracy.