Binary classification of spoken words with passive phononic metamaterials
This work addresses the problem of high energy consumption in AI by proposing a low-power computational substrate for always-on devices, though it is incremental as it focuses on a specific binary classification task.
The researchers tackled the challenge of using phononic metamaterials for machine learning by designing a non-periodic metamaterial from data samples, which successfully distinguished between pairs of spoken words with a simple readout nonlinearity, demonstrating their potential for zero-power smart devices.
Mitigating the energy requirements of artificial intelligence requires novel physical substrates for computation. Phononic metamaterials have a vanishingly low power dissipation and hence are a prime candidate for green, always-on computers. However, their use in machine learning applications has not been explored due to the complexity of their design process: Current phononic metamaterials are restricted to simple geometries (e.g. periodic, tapered), and hence do not possess sufficient expressivity to encode machine learning tasks. We design and fabricate a non-periodic phononic metamaterial, directly from data samples, that can distinguish between pairs of spoken words in the presence of a simple readout nonlinearity; hence demonstrating that phononic metamaterials are a viable avenue towards zero-power smart devices.