A dual-domain compute-in-memory system for general neural network inference
Related Articles
Real-time inference for binary neutron star mergers using machine learning
Mergers of binary neutron stars emit signals in both the gravitational-wave (GW) and electromagnetic spectra. Famously, the 2017 multi-messenger observation of GW170817 (refs. 1,2) led to scientific discoveries across cosmology3, nuclear physics4,5,6 and gravity7. Central to these results were the sky localization and distance obtained from the GW data, which, in the case of GW170817, helped to identify the associated electromagnetic transient, AT 2017gfo (ref. 8), 11 h after the GW signal. Fast analysis of GW data is critical for directing time-sensitive electromagnetic observations. However, owing to challenges arising from the length and complexity of signals, it is often necessary to make approximations that sacrifice accuracy. Here we present a machine-learning framework that performs complete binary neutron star inference in just 1 s without making any such approximations. Our approach enhances multi-messenger observations by providing: (1) accurate localization even before the merger; (2) improved localization precision by around 30% compared to approximate low-latency methods; and (3) detailed information on luminosity distance, inclination and masses, which can be used to prioritize expensive telescope time. Additionally, the flexibility and reduced cost of our method open new opportunities for equation-of-state studies. Finally, we demonstrate that our method scales to long signals, up to an hour in length, thus serving as a blueprint for data analysis for next-generation ground- and space-based detectors.
Innovating beyond electrophysiology through multimodal neural interfaces
Neural circuits distributed across different brain regions mediate how neural information is processed and integrated, resulting in complex cognitive capabilities and behaviour. To understand dynamics and interactions of neural circuits, it is crucial to capture the complete spectrum of neural activity, ranging from the fast action potentials of individual neurons to the population dynamics driven by slow brain-wide oscillations. In this Review, we discuss how advances in electrical and optical recording technologies, coupled with the emergence of machine learning methodologies, present a unique opportunity to unravel the complex dynamics of the brain. Although great progress has been made in both electrical and optical neural recording technologies, these alone fail to provide a comprehensive picture of the neuronal activity with high spatiotemporal resolution. To address this challenge, multimodal experiments integrating the complementary advantages of different techniques hold great promise. However, they are still hindered by the absence of multimodal data analysis methods capable of providing unified and interpretable explanations of the complex neural dynamics distinctly encoded in these modalities. Combining multimodal studies with advanced data analysis methods will offer novel perspectives to address unresolved questions in basic neuroscience and to develop treatments for various neurological disorders.
Error-driven upregulation of memory representations
Learning an association does not always succeed on the first attempt. Previous studies associated increased error signals in posterior medial frontal cortex with improved memory formation. However, the neurophysiological mechanisms that facilitate post-error learning remain poorly understood. To address this gap, participants performed a feedback-based association learning task and a 1-back localizer task. Increased hemodynamic responses in posterior medial frontal cortex were found for internal and external origins of memory error evidence, and during post-error encoding success as quantified by subsequent recall of face-associated memories. A localizer-based machine learning model displayed a network of cognitive control regions, including posterior medial frontal and dorsolateral prefrontal cortices, whose activity was related to face-processing evidence in the fusiform face area. Representation strength was higher during failed recall and increased during encoding when subsequent recall succeeded. These data enhance our understanding of the neurophysiological mechanisms of adaptive learning by linking the need for learning with increased processing of the relevant stimulus category.
A transient memory lapse in humans 1–3 h after training
In many non-human species, learning retention decreases temporarily following training. This has led to the suggestion that these lapses reflect a fundamental component of memory formation. If so, transient memory lapses should also be prevalent in humans, and should occur for all types of learning. In line with these predictions, we report two cases of transient memory lapses in humans that occur 1–3 h after training on a perceptual-discrimination task. The results indicate that the occurrence of transient memory lapses extends to perceptual learning, a form of skill learning, and suggest that transient memory lapses may be a common but overlooked facet of memory formation in humans.
Advancements in 2D layered material memristors: unleashing their potential beyond memory
The scalability of two-dimensional (2D) materials down to a single monolayer offers exciting prospects for high-speed, energy-efficient, scalable memristors. This review highlights the development of 2D material-based memristors and potential applications beyond memory, including neuromorphic, in-memory, in-sensor, and complex computing. This review also encompasses potential challenges and future opportunities for advancing these materials and technologies, underscoring the transformative impact of 2D memristors on versatile and sustainable electronic devices and systems.
Responses