The possibility to tune the Dzyaloshinskii Moriya interaction (DMI) by electric (E) field gating in ultra-thin magnetic materials has opened new perspectives in terms of controlling the stabilization of chiral spin structures. Most recent efforts have used voltage-induced charge redistribution at the interface between a metal and an oxide to modulate DMI. This approach is attractive for active devices but it tends to be volatile, making it energy demanding, and it is limited by Coulomb screening in the metal. Here we have demonstrated the non-volatile E-field manipulation of DMI by ionic liquid gating of Pt/Co/HfO2 ultra-thin films. The E-field effect on DMI scales with the E-field exposure time and is proposed to be linked to the migration and subsequent anchoring of oxygen species from the HfO2 layer into the Co and Pt layers. This effect permanently changes the properties of the material showing that E-fields can not only be used for local gating in devices but also as a highly scalable materials design tool for post-growth tuning of DMI.

While deep neural networks have surpassed human performance in multiple situations, they are prone to catastrophic forgetting: upon training a new task, they rapidly forget previously learned ones. Neuroscience studies, based on idealized tasks, suggest that in the brain, synapses overcome this issue by adjusting their plasticity depending on their past history. However, such “metaplastic” behaviors do not transfer directly to mitigate catastrophic forgetting in deep neural networks. In this work, we interpret the hidden weights used by binarized neural networks, a low-precision version of deep neural networks, as metaplastic variables, and modify their training technique to alleviate forgetting. Building on this idea, we propose and demonstrate experimentally, in situations of multitask and stream learning, a training technique that reduces catastrophic forgetting without needing previously presented data, nor formal boundaries between datasets and with performance approaching more mainstream techniques with task boundaries. We support our approach with a theoretical analysis on a tractable task. This work bridges computational neuroscience and deep learning, and presents significant assets for future embedded and neuromorphic systems, especially when using novel nanodevices featuring physics analogous to metaplasticity.