Advancing On-Device Neural Network Training with TinyPropv2: Dynamic, Sparse, and Efficient Backpropagation
This addresses efficient training for low-power microcontroller units in IoT applications, representing an incremental improvement over existing sparse methods.
The paper tackles on-device neural network training by introducing TinyPropv2, a dynamic sparse backpropagation algorithm that reduces computational effort to as little as 10% of full training while maintaining near-parity accuracy, with average drops of around 1% across datasets like CIFAR-10 and CIFAR-100.
This study introduces TinyPropv2, an innovative algorithm optimized for on-device learning in deep neural networks, specifically designed for low-power microcontroller units. TinyPropv2 refines sparse backpropagation by dynamically adjusting the level of sparsity, including the ability to selectively skip training steps. This feature significantly lowers computational effort without substantially compromising accuracy. Our comprehensive evaluation across diverse datasets CIFAR 10, CIFAR100, Flower, Food, Speech Command, MNIST, HAR, and DCASE2020 reveals that TinyPropv2 achieves near-parity with full training methods, with an average accuracy drop of only around 1 percent in most cases. For instance, against full training, TinyPropv2's accuracy drop is minimal, for example, only 0.82 percent on CIFAR 10 and 1.07 percent on CIFAR100. In terms of computational effort, TinyPropv2 shows a marked reduction, requiring as little as 10 percent of the computational effort needed for full training in some scenarios, and consistently outperforms other sparse training methodologies. These findings underscore TinyPropv2's capacity to efficiently manage computational resources while maintaining high accuracy, positioning it as an advantageous solution for advanced embedded device applications in the IoT ecosystem.