Fast Classification Learning with Neural Networks and Conceptors for Speech Recognition and Car Driving Maneuvers
This work addresses efficiency in training for speech recognition and autonomous driving applications, but appears incremental as it builds on existing methods with application-specific tweaks.
The paper tackles the problem of slow training in deep learning by using recurrent neural networks with conceptors to achieve fast learning with few examples, demonstrating improvements in speech recognition and car driving maneuver detection.
Recurrent neural networks are a powerful means in diverse applications. We show that, together with so-called conceptors, they also allow fast learning, in contrast to other deep learning methods. In addition, a relatively small number of examples suffices to train neural networks with high accuracy. We demonstrate this with two applications, namely speech recognition and detecting car driving maneuvers. We improve the state of the art by application-specific preparation techniques: For speech recognition, we use mel frequency cepstral coefficients leading to a compact representation of the frequency spectra, and detecting car driving maneuvers can be done without the commonly used polynomial interpolation, as our evaluation suggests.