Search for: All records

This study emulates associative learning in rodents by using a neuromorphic robot navigating an open-field arena. The goal is to investigate how biologically inspired neural models can reproduce animal-like learning behaviors in real-world robotic systems. We constructed a neuromorphic robot by deploying computational models of spatial and sensory neurons onto a mobile platform. Different coding schemes—rate coding for vibration signals and population coding for visual signals—were implemented. The associative learning model employs 19 spiking neurons and follows Hebbian plasticity principles to associate visual cues with favorable or unfavorable locations. Our robot successfully replicated classical rodent associative learning behavior by memorizing causal relationships between environmental cues and spatial outcomes. The robot’s self-learning capability emerged from repeated exposure and synaptic weight adaptation, without the need for labeled training data. Experiments confirmed functional learning behavior across multiple trials. This work provides a novel embodied platform for memory and learning research beyond traditional animal models. By embedding biologically inspired learning mechanisms into a real robot, we demonstrate how spatial memory can be formed and expressed through sensorimotor interactions. The model’s compact structure (19 neurons) illustrates a minimal yet functional learning network, and the study outlines principles for synaptic weight and threshold design, guiding future development of more complex neuromorphic systems. 

Deep neural networks (DNNs) have achieved remarkable success in various cognitive tasks through training on extensive labeled datasets. However, the heavy reliance on these datasets poses challenges for DNNs in scenarios with energy constraints in particular scenarios, such as on the moon. On the contrary, animals exhibit a self-learning capability by interacting with their surroundings and memorizing concurrent events without annotated data—a process known as associative learning. A classic example of associative learning is when a rat memorizes desired and undesired stimuli while exploring a T-maze. The successful implementation of associative learning aims to replicate the self-learning mechanisms observed in animals, addressing challenges in data-constrained environments. While current implementations of associative learning are predominantly small scale and offline, this work pioneers associative learning in a robot equipped with a neuromorphic chip, specifically for online learning in a T-maze. The system successfully replicates classic associative learning observed in rodents, using neuromorphic robots as substitutes for rodents. The neuromorphic robot autonomously learns the cause-and-effect relationship between audio and visual stimuli. 

Federated learning is a novel paradigm allowing the training of a global machine-learning model on distributed devices. It shares model parameters instead of private raw data during the entire model training process. While federated learning enables machine learning processes to take place collaboratively on Internet of Things (IoT) devices, compared to data centers, IoT devices with limited resource budgets typically have less security protection and are more vulnerable to potential thermal stress. Current research on the evaluation of federated learning is mainly based on the simulation of multi-clients/processes on a single machine/device. However, there is a gap in understanding the performance of federated learning under thermal stress in real-world distributed low-power heterogeneous IoT devices. Our previous work was among the first to evaluate the performance of federated learning under thermal stress on real-world IoT-based distributed systems. In this paper, we extended our work to a larger scale of heterogeneous real-world IoT-based distributed systems to further evaluate the performance of federated learning under thermal stress. To the best of our knowledge, the presented work is among the first to evaluate the performance of federated learning under thermal stress on real-world heterogeneous IoT-based systems. We conducted comprehensive experiments using the MNIST dataset and various performance metrics, including training time, CPU and GPU utilization rate, temperature, and power consumption. We varied the proportion of clients under thermal stress in each group of experiments and systematically quantified the effectiveness and real-world impact of thermal stress on the low-end heterogeneous IoT-based federated learning system. We added 67% more training epochs and 50% more clients compared with our previous work. The experimental results demonstrate that thermal stress is still effective on IoT-based federated learning systems as the entire global model and device performance degrade when even a small ratio of IoT devices are being impacted. Experimental results have also shown that the more influenced client under thermal stress within the federated learning system (FLS) tends to have a more major impact on the performance of FLS under thermal stress. 

Abstract Mid‐lithosphere discontinuities are seismic interfaces likely located within the lithospheric mantle of stable cratons, which typically represent velocities decreasing with depth. The origins of these interfaces are poorly understood due to the difficulties in both characterizing them seismically and reconciling the observations with thermal‐chemical models of cratons. Metasomatism of the cratonic lithosphere has been reported by numerous geochemical and petrological studies worldwide, yet its seismic signature remains elusive. Here, we identify two distinct mid‐lithosphere discontinuities at ∼87 and ∼117 km depth beneath the eastern Wyoming craton and the southwestern Superior craton by analyzing seismic data recorded by two longstanding stations. Our waveform modeling shows that the shallow and deep interfaces represent isotropic velocity drops of 2%–8% and 4%–9%, respectively, depending on the contributions from changes in radial anisotropy and density. By building a thermal‐chemical model including the regional xenolith thermobarometry constraints and the experimental phase‐equilibrium data of mantle metasomatism, we show that the shallow interface probably represents the metasomatic front, below which hydrous minerals such as amphibole and phlogopite are present, whereas the deep interface may be caused by the onset of carbonated partial melting. The hydrous minerals and melts are products of mantle metasomatism, with CO₂‐H₂O‐rich siliceous melt as a probable metasomatic reagent. Our results suggest that mantle metasomatism is probably an important cause of mid‐lithosphere discontinuities worldwide, especially near craton boundaries, where the mantle lithosphere may be intensely metasomatized by fluids and melts released by subducting slabs. 

Abstract Alaska is a tectonically active region with a long history of subduction and terrane accretion, but knowledge of its deep seismic structure is limited by a relatively sparse station distribution. By combining data from the EarthScope Transportable Array and other regional seismic networks, we obtain a high‐resolution state‐wide map of the Moho and upper‐mantle discontinuities beneath Alaska using teleseismic SH‐wave reverberations. Crustal thickness is generally correlated with elevation and the deepest Moho is in the region with basal accretion of the subducted Yakutat plate, consistent with its higher density due to a more mafic composition. The crustal thickness in the Brooks Range agrees with the prediction based on Airy isostasy and the weak free‐air gravity anomaly, suggesting that this region probably does not have significant density anomalies. We also resolve the 410, 520, and 660 discontinuities in most regions, with a thickened mantle transition zone (MTZ) and a normal depth difference between the 520 and 660 discontinuities (d660‐d520) under central Alaska, indicating the presence of the subducted Pacific slab in the upper MTZ. A near‐normal MTZ and a significantly smaller d660‐d520 are resolved under southeastern Alaska, suggesting potential mantle upwelling in the lower MTZ. Beneath the Alaska Peninsula, the thinned MTZ implies that the Pacific slab may not have reached the MTZ in this region, which is also consistent with recent tomography models. Overall, the results demonstrate a bent or segmented Pacific slab with varying depths under central Alaska and the Alaska Peninsula. 

Abstract Seismic noise has been widely used to image Earth's structure in the past decades as a powerful supplement to earthquake signals. Although the seismic noise field contains both surface‐wave and body‐wave components, most previous studies have focused on surface waves due to their large amplitudes. Here, we use array analyses to identify body‐wave noise traveling asPKPwaves. We find that by cross‐correlating the array‐stacked horizontal‐ and vertical‐component data in the time windows containing thePKPnoise signals, we extract a phase likely representingPKS‐PKP, the differential phase betweenPKSandPKP. This phase can potentially be used for shear‐wave‐splitting analysis. Our results also suggest that the sources of body‐wave noise are extremely heterogeneous in both space and time, which should be accounted for in future studies using body‐wave noise to image Earth structure. 

Abstract Ocean transform faults often generate characteristic earthquakes that repeatedly rupture the same fault patches. The westernmost Gofar transform fault quasi‐periodically hosts ∼M6 earthquakes every ∼5 years, and microseismicity suggests that the fault is segmented into five distinct zones, including a rupture barrier zone that may have modulated the rupture of adjacentM6 earthquakes. However, the relationship between the systematic slip behavior of the Gofar fault and the fault material properties is still poorly known. Specifically, the role of pore fluids in regulating the slip of the Gofar fault is unclear. Here, we use differential travel times between nearby earthquakes to estimate the in‐situV_p/V_sof the fault‐zone materials. We apply this technique to the dataset collected by an ocean‐bottom‐seismometer network deployed around the Gofar fault in 2008, which recorded abundant microearthquakes, and find a moderateV_p/V_sof 1.75–1.80 in the rupture barrier zone and a lowV_p/V_sof 1.61–1.69 in the down‐dip edge of the 2008M6 rupture zone. This lateral variation inV_p/V_smay be caused by both pore fluids and chemical alteration. We also find a 5%–10% increase inV_p/V_sin the barrier zone during the 9 months before the mainshock. This increase may have been caused by fluid migrations or slip transients in the barrier zone. 

Abstract Lithospheric discontinuities, including the lithosphere‐asthenosphere boundary (LAB) and the enigmatic mid‐lithospheric discontinuities (MLDs), hold important clues about the structure and evolution of tectonic plates. However, P‐ and S‐receiver‐function (PRF and SRF) techniques, two traditional techniques to image Earth's deep discontinuities, have some shortcomings in imaging lithosphere discontinuities. Here, we propose a new method using reflections generated by teleseismic S waves (hereafter S‐reflections) to image lithospheric discontinuities, which are less affected by multiple phases than PRFs and have better depth resolution than SRFs. We apply this method to the data collected by the Transportable Array and other regional seismic networks and obtain new high‐resolution images of the lithosphere below the contiguous US. Beneath the tectonically active Western US, we observe a negative polarity reflector (NPR) in the depth range of 60–110 km, with greatly varying amplitude and depth, which correlates with active tectonic processes. We interpret this feature as the LAB below the Western US. Beneath the tectonically stable Central and Eastern US, we observe two NPRs in the depth ranges of 60–100 km and 100–150 km, whose amplitude and depth also vary significantly, and which appear to correlate with past tectonic processes. We interpret these features as MLDs below the Central and Eastern US. Our results show reasonable agreement with results from PRFs, which have similar depth resolution, suggesting the possibility of joint inversion of S‐reflections and PRFs to constrain the properties of lithospheric discontinuities.