Pub Date : 2022-10-19DOI: 10.1088/2634-4386/ac9b85
Corentin Scholaert, Kamila Janzakova, Y. Coffinier, F. Alibart, Sébastien Pecqueur
The brain capitalizes on the complexity of both its biochemistry for neurons to encode diverse pieces of information with various neurotransmitters and its morphology at multiple scales to route different pathways for neural interconnectivity. Conducting polymer dendrites can show similar features by differentiating between cations and anions thanks to their charge accumulation profile and the asymmetry in their dendriticity that allows projecting spike signals differently. Here, we exploit such mimicry for in materio classification of bursting activity and investigate, in phosphate buffered saline, the capability of such object to sense bursts of voltage pulses of 100 mV amplitude, emitted by a local gate in the vicinity of the dendrite. The dendrite integrates the different activities with a fading memory time window that is characteristic of both the polarity of the spikes and the temporality of the burst. By this first demonstration, the ‘material-object’ definitely shows great potential to be a node halfway between the two realms of brain and electronic communication.
{"title":"Plasticity of conducting polymer dendrites to bursts of voltage spikes in phosphate buffered saline","authors":"Corentin Scholaert, Kamila Janzakova, Y. Coffinier, F. Alibart, Sébastien Pecqueur","doi":"10.1088/2634-4386/ac9b85","DOIUrl":"https://doi.org/10.1088/2634-4386/ac9b85","url":null,"abstract":"The brain capitalizes on the complexity of both its biochemistry for neurons to encode diverse pieces of information with various neurotransmitters and its morphology at multiple scales to route different pathways for neural interconnectivity. Conducting polymer dendrites can show similar features by differentiating between cations and anions thanks to their charge accumulation profile and the asymmetry in their dendriticity that allows projecting spike signals differently. Here, we exploit such mimicry for in materio classification of bursting activity and investigate, in phosphate buffered saline, the capability of such object to sense bursts of voltage pulses of 100 mV amplitude, emitted by a local gate in the vicinity of the dendrite. The dendrite integrates the different activities with a fading memory time window that is characteristic of both the polarity of the spikes and the temporality of the burst. By this first demonstration, the ‘material-object’ definitely shows great potential to be a node halfway between the two realms of brain and electronic communication.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123876333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-09DOI: 10.1088/2634-4386/ac9899
Xiaocong Du, S. Venkataramanaiah, Zheng Li, Han-Sok Suh, Shihui Yin, Gokul Krishnan, Frank Liu, Jae-sun Seo, Yu Cao
There is an increasing need for continual learning in dynamic systems at the edge, such as self-driving vehicles, surveillance drones, and robotic systems. Such a system requires learning from the data stream, training the model to preserve previous information and adapt to a new task, and generating a single-headed vector for future inference, within a limited power budget. Different from previous continual learning algorithms with dynamic structures, this work focuses on a single network and model segmentation to mitigate catastrophic forgetting problem. Leveraging the redundant capacity of a single network, model parameters for each task are separated into two groups: one important group which is frozen to preserve current knowledge, and a secondary group to be saved (not pruned) for future learning. A fixed-size memory containing a small amount of previously seen data is further adopted to assist the training. Without additional regularization, the simple yet effective approach of progressive segmented training (PST) successfully incorporates multiple tasks and achieves state-of-the-art accuracy in the single-head evaluation on the CIFAR-10 and CIFAR-100 datasets. Moreover, the segmented training significantly improves computation efficiency in continual learning and thus, enabling efficient continual learning at the edge. On Intel Stratix-10 MX FPGA, we further demonstrate the efficiency of PST with representative CNNs trained on CIFAR-10.
{"title":"Efficient continual learning at the edge with progressive segmented training","authors":"Xiaocong Du, S. Venkataramanaiah, Zheng Li, Han-Sok Suh, Shihui Yin, Gokul Krishnan, Frank Liu, Jae-sun Seo, Yu Cao","doi":"10.1088/2634-4386/ac9899","DOIUrl":"https://doi.org/10.1088/2634-4386/ac9899","url":null,"abstract":"There is an increasing need for continual learning in dynamic systems at the edge, such as self-driving vehicles, surveillance drones, and robotic systems. Such a system requires learning from the data stream, training the model to preserve previous information and adapt to a new task, and generating a single-headed vector for future inference, within a limited power budget. Different from previous continual learning algorithms with dynamic structures, this work focuses on a single network and model segmentation to mitigate catastrophic forgetting problem. Leveraging the redundant capacity of a single network, model parameters for each task are separated into two groups: one important group which is frozen to preserve current knowledge, and a secondary group to be saved (not pruned) for future learning. A fixed-size memory containing a small amount of previously seen data is further adopted to assist the training. Without additional regularization, the simple yet effective approach of progressive segmented training (PST) successfully incorporates multiple tasks and achieves state-of-the-art accuracy in the single-head evaluation on the CIFAR-10 and CIFAR-100 datasets. Moreover, the segmented training significantly improves computation efficiency in continual learning and thus, enabling efficient continual learning at the edge. On Intel Stratix-10 MX FPGA, we further demonstrate the efficiency of PST with representative CNNs trained on CIFAR-10.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117083062","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-03DOI: 10.1088/2634-4386/ac970c
Yingfen Wei, G. Vats, B. Noheda
The discovery of ferroelectricity in HfO2-based thin films brings tremendous opportunities for emerging ferroelectric memories as well as for synaptic devices. The origin of ferroelectricity in this material is widely attributed to the presence of a polar orthorhombic phase. However, a new ferroelectric rhombohedral phase displaying large polarization with no need of pre-cycling, has more recently been reported in epitaxial Hf0.5Zr0.5O2 (HZO). In this work, the switching mechanism of the rhombohedral phase of HZO films is characterized by a two-stage process. In addition, the synaptic behaviour of this phase is presented, comparing it with previous reports on orthorhombic or non-epitaxial films. Unexpected similarities have been found between these structurally distinct systems. Even though the epitaxial films present a larger coercive field, the ration between the activation field for intrinsic polarization switching and the coercive field (F a/E c) has been found to be close to 2, in agreement with that reported for other hafnia samples. This is about 5 times smaller than in most other ferroelectrics, confirming this characteristic as a unique feature of hafnia-based ferroelectrics.
{"title":"Synaptic behaviour in ferroelectric epitaxial rhombohedral Hf0.5Zr0.5O2 thin films","authors":"Yingfen Wei, G. Vats, B. Noheda","doi":"10.1088/2634-4386/ac970c","DOIUrl":"https://doi.org/10.1088/2634-4386/ac970c","url":null,"abstract":"The discovery of ferroelectricity in HfO2-based thin films brings tremendous opportunities for emerging ferroelectric memories as well as for synaptic devices. The origin of ferroelectricity in this material is widely attributed to the presence of a polar orthorhombic phase. However, a new ferroelectric rhombohedral phase displaying large polarization with no need of pre-cycling, has more recently been reported in epitaxial Hf0.5Zr0.5O2 (HZO). In this work, the switching mechanism of the rhombohedral phase of HZO films is characterized by a two-stage process. In addition, the synaptic behaviour of this phase is presented, comparing it with previous reports on orthorhombic or non-epitaxial films. Unexpected similarities have been found between these structurally distinct systems. Even though the epitaxial films present a larger coercive field, the ration between the activation field for intrinsic polarization switching and the coercive field (F a/E c) has been found to be close to 2, in agreement with that reported for other hafnia samples. This is about 5 times smaller than in most other ferroelectrics, confirming this characteristic as a unique feature of hafnia-based ferroelectrics.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123376793","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-22DOI: 10.1088/2634-4386/acb841
Ismael Ribeiro de Assis, I. Mertig, B. Göbel
Magnetic skyrmions are nanoscale magnetic whirls that are highly stable and can be moved by currents. They have led to the prediction of a skyrmion-based artificial neuron device with leak-integrate-fire functionality. However, so far, these devices lack a refractory process, estimated to be crucial for neuronal dynamics. Here we demonstrate that a biskyrmion-based artificial neuron overcomes this insufficiency. When driven by spin-orbit torques, a single biskyrmion splits into two subskyrmions that move towards a designated location and can be detected electrically, ultimately resembling the excitation process of a neuron that fires. The attractive interaction of the two skyrmions leads to a unique trajectory: Once they reach the detector area, they automatically return to the center to reform the biskyrmion but on a different path. During this reset period, the neuron cannot fire again. Our suggested device resembles a biological neuron with the leak, integrate, fire and refractory characteristics increasing the bio-fidelity of current skyrmion-based devices.
{"title":"Biskyrmion-based artificial neuron","authors":"Ismael Ribeiro de Assis, I. Mertig, B. Göbel","doi":"10.1088/2634-4386/acb841","DOIUrl":"https://doi.org/10.1088/2634-4386/acb841","url":null,"abstract":"Magnetic skyrmions are nanoscale magnetic whirls that are highly stable and can be moved by currents. They have led to the prediction of a skyrmion-based artificial neuron device with leak-integrate-fire functionality. However, so far, these devices lack a refractory process, estimated to be crucial for neuronal dynamics. Here we demonstrate that a biskyrmion-based artificial neuron overcomes this insufficiency. When driven by spin-orbit torques, a single biskyrmion splits into two subskyrmions that move towards a designated location and can be detected electrically, ultimately resembling the excitation process of a neuron that fires. The attractive interaction of the two skyrmions leads to a unique trajectory: Once they reach the detector area, they automatically return to the center to reform the biskyrmion but on a different path. During this reset period, the neuron cannot fire again. Our suggested device resembles a biological neuron with the leak, integrate, fire and refractory characteristics increasing the bio-fidelity of current skyrmion-based devices.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131789042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-13DOI: 10.1088/2634-4386/ac91a0
Z. Kuncic, T. Nakayama, J. Gimzewski
This NCE Focus Issue is motivated by the intriguingly neuromorphic properties of many-body systems self-assembled from nanoscale elementary components. The rationale behind this is that biological neural networks, including in particular their nanoscale synapses, are formed by bottom-up self-assembly, rather than top-down design. Self-assembled nanosystems inherit a disordered network structure and the nonlinear interactions between the networked elements can give rise to emergent properties, as espoused by the legendary Nobel laureate Phillip W. Anderson in his famous article “More is Different” (Science 177, 393, 1972).
这个NCE焦点问题的动机是由纳米级基本组件自组装的多体系统的有趣的神经形态特性。这背后的基本原理是，生物神经网络，特别是它们的纳米级突触，是由自下而上的自组装形成的，而不是自上而下的设计。自组装纳米系统继承了无序的网络结构，网络元素之间的非线性相互作用可以产生涌现特性，正如诺贝尔奖传奇获得者Phillip W. Anderson在他著名的文章“More is Different”(Science 177, 393, 1972)中所支持的那样。
{"title":"Editorial: Focus on disordered, self-assembled neuromorphic systems","authors":"Z. Kuncic, T. Nakayama, J. Gimzewski","doi":"10.1088/2634-4386/ac91a0","DOIUrl":"https://doi.org/10.1088/2634-4386/ac91a0","url":null,"abstract":"\u0000 This NCE Focus Issue is motivated by the intriguingly neuromorphic properties of many-body systems self-assembled from nanoscale elementary components. The rationale behind this is that biological neural networks, including in particular their nanoscale synapses, are formed by bottom-up self-assembly, rather than top-down design. Self-assembled nanosystems inherit a disordered network structure and the nonlinear interactions between the networked elements can give rise to emergent properties, as espoused by the legendary Nobel laureate Phillip W. Anderson in his famous article “More is Different” (Science 177, 393, 1972).","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114619317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-07DOI: 10.1088/2634-4386/ac9012
S. Brivio, S. Spiga, D. Ielmini
HfO2-based resistive switching memory (RRAM) combines several outstanding properties, such as high scalability, fast switching speed, low power, compatibility with complementary metal-oxide-semiconductor technology, with possible high-density or three-dimensional integration. Therefore, today, HfO2 RRAMs have attracted a strong interest for applications in neuromorphic engineering, in particular for the development of artificial synapses in neural networks. This review provides an overview of the structure, the properties and the applications of HfO2-based RRAM in neuromorphic computing. Both widely investigated applications of nonvolatile devices and pioneering works about volatile devices are reviewed. The RRAM device is first introduced, describing the switching mechanisms associated to filamentary path of HfO2 defects such as oxygen vacancies. The RRAM programming algorithms are described for high-precision multilevel operation, analog weight update in synaptic applications and for exploiting the resistance dynamics of volatile devices. Finally, the neuromorphic applications are presented, illustrating both artificial neural networks with supervised training and with multilevel, binary or stochastic weights. Spiking neural networks are then presented for applications ranging from unsupervised training to spatio-temporal recognition. From this overview, HfO2-based RRAM appears as a mature technology for a broad range of neuromorphic computing systems.
{"title":"HfO2-based resistive switching memory devices for neuromorphic computing","authors":"S. Brivio, S. Spiga, D. Ielmini","doi":"10.1088/2634-4386/ac9012","DOIUrl":"https://doi.org/10.1088/2634-4386/ac9012","url":null,"abstract":"HfO2-based resistive switching memory (RRAM) combines several outstanding properties, such as high scalability, fast switching speed, low power, compatibility with complementary metal-oxide-semiconductor technology, with possible high-density or three-dimensional integration. Therefore, today, HfO2 RRAMs have attracted a strong interest for applications in neuromorphic engineering, in particular for the development of artificial synapses in neural networks. This review provides an overview of the structure, the properties and the applications of HfO2-based RRAM in neuromorphic computing. Both widely investigated applications of nonvolatile devices and pioneering works about volatile devices are reviewed. The RRAM device is first introduced, describing the switching mechanisms associated to filamentary path of HfO2 defects such as oxygen vacancies. The RRAM programming algorithms are described for high-precision multilevel operation, analog weight update in synaptic applications and for exploiting the resistance dynamics of volatile devices. Finally, the neuromorphic applications are presented, illustrating both artificial neural networks with supervised training and with multilevel, binary or stochastic weights. Spiking neural networks are then presented for applications ranging from unsupervised training to spatio-temporal recognition. From this overview, HfO2-based RRAM appears as a mature technology for a broad range of neuromorphic computing systems.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116739439","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-06DOI: 10.1088/2634-4386/acc204
Ioannis E. Polykretis, Lazar Supic, A. Danielescu
Beyond providing accurate movements, achieving smooth motion trajectories is a long-standing goal of robotics control theory for arms aiming to replicate natural human movements. Drawing inspiration from biological agents, whose reaching control networks effortlessly give rise to smooth and precise movements, can simplify these control objectives for robot arms. Neuromorphic processors, which mimic the brain’s computational principles, are an ideal platform to approximate the accuracy and smoothness of biological controllers while maximizing their energy efficiency and robustness. However, the incompatibility of conventional control methods with neuromorphic hardware limits the computational efficiency and explainability of their existing adaptations. In contrast, the neuronal subnetworks underlying smooth and accurate reaching movements are effective, minimal, and inherently compatible with neuromorphic hardware. In this work, we emulate these networks with a biologically realistic spiking neural network for motor control on neuromorphic hardware. The proposed controller incorporates experimentally-identified short-term synaptic plasticity and specialized neurons that regulate sensory feedback gain to provide smooth and accurate joint control across a wide motion range. Concurrently, it preserves the minimal complexity of its biological counterpart and is directly deployable on Intel’s neuromorphic processor. Using the joint controller as a building block and inspired by joint coordination in human arms, we scaled up this approach to control real-world robot arms. The trajectories and smooth, bell-shaped velocity profiles of the resulting motions resembled those of humans, verifying the biological relevance of the controller. Notably, the method achieved state-of-the-art control performance while decreasing the motion jerk by 19% to improve motion smoothness. Overall, this work suggests that control solutions inspired by experimentally identified neuronal architectures can provide effective, neuromorphic-controlled robots.
{"title":"Bioinspired smooth neuromorphic control for robotic arms","authors":"Ioannis E. Polykretis, Lazar Supic, A. Danielescu","doi":"10.1088/2634-4386/acc204","DOIUrl":"https://doi.org/10.1088/2634-4386/acc204","url":null,"abstract":"Beyond providing accurate movements, achieving smooth motion trajectories is a long-standing goal of robotics control theory for arms aiming to replicate natural human movements. Drawing inspiration from biological agents, whose reaching control networks effortlessly give rise to smooth and precise movements, can simplify these control objectives for robot arms. Neuromorphic processors, which mimic the brain’s computational principles, are an ideal platform to approximate the accuracy and smoothness of biological controllers while maximizing their energy efficiency and robustness. However, the incompatibility of conventional control methods with neuromorphic hardware limits the computational efficiency and explainability of their existing adaptations. In contrast, the neuronal subnetworks underlying smooth and accurate reaching movements are effective, minimal, and inherently compatible with neuromorphic hardware. In this work, we emulate these networks with a biologically realistic spiking neural network for motor control on neuromorphic hardware. The proposed controller incorporates experimentally-identified short-term synaptic plasticity and specialized neurons that regulate sensory feedback gain to provide smooth and accurate joint control across a wide motion range. Concurrently, it preserves the minimal complexity of its biological counterpart and is directly deployable on Intel’s neuromorphic processor. Using the joint controller as a building block and inspired by joint coordination in human arms, we scaled up this approach to control real-world robot arms. The trajectories and smooth, bell-shaped velocity profiles of the resulting motions resembled those of humans, verifying the biological relevance of the controller. Notably, the method achieved state-of-the-art control performance while decreasing the motion jerk by 19% to improve motion smoothness. Overall, this work suggests that control solutions inspired by experimentally identified neuronal architectures can provide effective, neuromorphic-controlled robots.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133877880","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-01DOI: 10.1088/2634-4386/ac8827
B. Shi, N. Calabretta, R. Stabile
We experimentally demonstrate the emulation of scaling of the semiconductor optical amplifier (SOA) based integrated all-optical neural network in terms of number of input channels and layer cascade, with chromatic input at the neuron and monochromatic output conversion, obtained by exploiting cross-gain-modulation effect. We propose a noise model for investigating the signal degradation on the signal processing after cascades of SOAs, and we validate it via experimental results. Both experiments and simulations claim that the all-optical neuron (AON), with wavelength conversion as non-linear function, is able to compress noise for noisy optical inputs. This suggests that the use of SOA-based AON with wavelength conversion may allow for building neural networks with arbitrary depth. In fact, an arbitrarily deep neural network, built out of seven-channel input AONs, is shown to guarantee an error minor than 0.1 when operating at input power levels of −20 dBm/channel and with a 6 dB input dynamic range. Then the simulations results, extended to an arbitrary number of input channels and layers, suggest that by cascading and interconnecting multiple of these monolithically integrated AONs, it is possible to build a neural network with 12-inputs/neuron 12 neurons/layer and arbitrary depth scaling, or an 18-inputs/neuron 18-neurons/layer for single layer implementation, to maintain an output error <0.1. Further improvement in height scalability can be obtained by optimizing the input power.
{"title":"Emulation and modelling of semiconductor optical amplifier-based all-optical photonic integrated deep neural network with arbitrary depth","authors":"B. Shi, N. Calabretta, R. Stabile","doi":"10.1088/2634-4386/ac8827","DOIUrl":"https://doi.org/10.1088/2634-4386/ac8827","url":null,"abstract":"We experimentally demonstrate the emulation of scaling of the semiconductor optical amplifier (SOA) based integrated all-optical neural network in terms of number of input channels and layer cascade, with chromatic input at the neuron and monochromatic output conversion, obtained by exploiting cross-gain-modulation effect. We propose a noise model for investigating the signal degradation on the signal processing after cascades of SOAs, and we validate it via experimental results. Both experiments and simulations claim that the all-optical neuron (AON), with wavelength conversion as non-linear function, is able to compress noise for noisy optical inputs. This suggests that the use of SOA-based AON with wavelength conversion may allow for building neural networks with arbitrary depth. In fact, an arbitrarily deep neural network, built out of seven-channel input AONs, is shown to guarantee an error minor than 0.1 when operating at input power levels of −20 dBm/channel and with a 6 dB input dynamic range. Then the simulations results, extended to an arbitrary number of input channels and layers, suggest that by cascading and interconnecting multiple of these monolithically integrated AONs, it is possible to build a neural network with 12-inputs/neuron 12 neurons/layer and arbitrary depth scaling, or an 18-inputs/neuron 18-neurons/layer for single layer implementation, to maintain an output error <0.1. Further improvement in height scalability can be obtained by optimizing the input power.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114877560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-01DOI: 10.1088/2634-4386/ac969a
Marie Drouhin, Shuaifei Li, M. Grelier, S. Collin, F. Godel, R. Elliman, B. Dlubak, J. Trastoy, D. Querlioz, J. Grollier
Hardware spiking neural networks hold the promise of realizing artificial intelligence with high energy efficiency. In this context, solid-state and scalable memristors can be used to mimic biological neuron characteristics. However, these devices show limited neuronal behaviors and have to be integrated in more complex circuits to implement the rich dynamics of biological neurons. Here we studied a NbO x memristor neuron that is capable of emulating numerous neuronal dynamics, including tonic spiking, stochastic spiking, leaky-integrate-and-fire features, spike latency, temporal integration. The device also exhibits phasic bursting, a property that has scarcely been observed and studied in solid-state nano-neurons. We show that we can reproduce and understand this particular response through simulations using non-linear dynamics. These results show that a single NbO x device is sufficient to emulate a collection of rich neuronal dynamics that paves a path forward for realizing scalable and energy-efficient neuromorphic computing paradigms.
{"title":"Characterization and modeling of spiking and bursting in experimental NbO x neuron","authors":"Marie Drouhin, Shuaifei Li, M. Grelier, S. Collin, F. Godel, R. Elliman, B. Dlubak, J. Trastoy, D. Querlioz, J. Grollier","doi":"10.1088/2634-4386/ac969a","DOIUrl":"https://doi.org/10.1088/2634-4386/ac969a","url":null,"abstract":"Hardware spiking neural networks hold the promise of realizing artificial intelligence with high energy efficiency. In this context, solid-state and scalable memristors can be used to mimic biological neuron characteristics. However, these devices show limited neuronal behaviors and have to be integrated in more complex circuits to implement the rich dynamics of biological neurons. Here we studied a NbO x memristor neuron that is capable of emulating numerous neuronal dynamics, including tonic spiking, stochastic spiking, leaky-integrate-and-fire features, spike latency, temporal integration. The device also exhibits phasic bursting, a property that has scarcely been observed and studied in solid-state nano-neurons. We show that we can reproduce and understand this particular response through simulations using non-linear dynamics. These results show that a single NbO x device is sufficient to emulate a collection of rich neuronal dynamics that paves a path forward for realizing scalable and energy-efficient neuromorphic computing paradigms.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128320227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-26DOI: 10.1088/2634-4386/acb37e
S. Neumayer, A. Ievlev, A. Tselev, S. Basun, B. Conner, M. Susner, P. Maksymovych
Smart electronic circuits that support neuromorphic computing on the hardware level necessitate materials with memristive, memcapacitive, and neuromorphic- like functional properties; in short, the electronic response must depend on the voltage history, thus enabling learning algorithms. Here we demonstrate volatile ferroelectric switching of Sn2P2S6 at room temperature and see that initial polarization orientation strongly determines the properties of polarization switching. In particular, polarization switching hysteresis is strongly imprinted by the original polarization state, shifting the regions of non-linearity toward zero-bias. As a corollary, polarization switching also enables effective capacitive switching, approaching the sought-after regime of memcapacitance. Landau–Ginzburg–Devonshire simulations demonstrate that one mechanism by which polarization can control the shape of the hysteresis loop is the existence of charged domain walls (DWs) decorating the periphery of the repolarization nucleus. These walls oppose the growth of the switched domain and favor back-switching, thus creating a scenario of controlled volatile ferroelectric switching. Although the measurements were carried out with single crystals, prospectively volatile polarization switching can be tuned by tailoring sample thickness, DW mobility and electric fields, paving way to non-linear dielectric properties for smart electronic circuits.
{"title":"Polarization-controlled volatile ferroelectric and capacitive switching in Sn2P2S6","authors":"S. Neumayer, A. Ievlev, A. Tselev, S. Basun, B. Conner, M. Susner, P. Maksymovych","doi":"10.1088/2634-4386/acb37e","DOIUrl":"https://doi.org/10.1088/2634-4386/acb37e","url":null,"abstract":"Smart electronic circuits that support neuromorphic computing on the hardware level necessitate materials with memristive, memcapacitive, and neuromorphic- like functional properties; in short, the electronic response must depend on the voltage history, thus enabling learning algorithms. Here we demonstrate volatile ferroelectric switching of Sn2P2S6 at room temperature and see that initial polarization orientation strongly determines the properties of polarization switching. In particular, polarization switching hysteresis is strongly imprinted by the original polarization state, shifting the regions of non-linearity toward zero-bias. As a corollary, polarization switching also enables effective capacitive switching, approaching the sought-after regime of memcapacitance. Landau–Ginzburg–Devonshire simulations demonstrate that one mechanism by which polarization can control the shape of the hysteresis loop is the existence of charged domain walls (DWs) decorating the periphery of the repolarization nucleus. These walls oppose the growth of the switched domain and favor back-switching, thus creating a scenario of controlled volatile ferroelectric switching. Although the measurements were carried out with single crystals, prospectively volatile polarization switching can be tuned by tailoring sample thickness, DW mobility and electric fields, paving way to non-linear dielectric properties for smart electronic circuits.","PeriodicalId":198030,"journal":{"name":"Neuromorphic Computing and Engineering","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127124429","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：