Academia.eduAcademia.edu

Figure 2 – uploaded by Arpit Shah

See full PDFdownloadDownload figure
Figure 2: A feedforward SNAN where an astrocyte senses and imposes synchronous neuronal activity. A. SNAN Architecture B. Astrocytic and neuronal activity on Loihi. Introducing Astrocytes on a Neuromorphic Processor

Figure 2 A feedforward SNAN where an astrocyte senses and imposes synchronous neuronal activity. A. SNAN Architecture B. Astrocytic and neuronal activity on Loihi. Introducing Astrocytes on a Neuromorphic Processor

Related Figures (7)

Figure 1: A. Schematic of the biological tripartite synapse. B. Its implementaton on our Loihi Astrocytic Module, preserv- ing the communcation pathways between the astrocytic and the neuronal components of the SNAN.
Figure 1: A. Schematic of the biological tripartite synapse. B. Its implementaton on our Loihi Astrocytic Module, preserv- ing the communcation pathways between the astrocytic and the neuronal components of the SNAN.
Figure 3: Overall Astrocyte SDK Structure. The Astrocyte SDK builds on the existing NxSDK Python-based programming model for Loihi [35] and it is designed to facilitate the integration of astrocytes with SNNs. The SDK’s API provides access to 3 main classes, depicted in Figure 3:
Figure 3: Overall Astrocyte SDK Structure. The Astrocyte SDK builds on the existing NxSDK Python-based programming model for Loihi [35] and it is designed to facilitate the integration of astrocytes with SNNs. The SDK’s API provides access to 3 main classes, depicted in Figure 3:
Figure 4: An example Astrocyte SDK program for integrat ing 2 astrocytes into an SNAN using the AstrocyteGroup.
Figure 4: An example Astrocyte SDK program for integrat ing 2 astrocytes into an SNAN using the AstrocyteGroup.
Figure 5: An astrocyte group inducing synchronized activi- ties in postsynaptic neurons.
Figure 5: An astrocyte group inducing synchronized activi- ties in postsynaptic neurons.
Figure 6: Single-shot pattern memorization and retrieval using an SNAN on Loihi. A. Architecture of the feedforward SNAN with astrocyte-reinforced STDP. B. Activity of the sensory neurons and the memory neuron during learning and retrieval. C. Dynamics of the synaptic weights potentiated by the astrocyte-reinforced STDP.
Figure 6: Single-shot pattern memorization and retrieval using an SNAN on Loihi. A. Architecture of the feedforward SNAN with astrocyte-reinforced STDP. B. Activity of the sensory neurons and the memory neuron during learning and retrieval. C. Dynamics of the synaptic weights potentiated by the astrocyte-reinforced STDP.
Figure 7: Homeostatic astrocyte on Loihi sensing transition from ordered to chaotic network state. A. SNAN architecture showing training and testing phases. B. Learned weights for all the neuronal-astrocytic connections. C. (upper row) Ordered and chaotic neuronal network state (middle row) number of active neurons in both states showing similar neuronal activity (bottom row) astrocytic activity during the transition from order to chaos. Introducing Astrocytes on a Neuromorphic Processor
Figure 7: Homeostatic astrocyte on Loihi sensing transition from ordered to chaotic network state. A. SNAN architecture showing training and testing phases. B. Learned weights for all the neuronal-astrocytic connections. C. (upper row) Ordered and chaotic neuronal network state (middle row) number of active neurons in both states showing similar neuronal activity (bottom row) astrocytic activity during the transition from order to chaos. Introducing Astrocytes on a Neuromorphic Processor
Input N; was represented as the integrated current of real-time presynaptic spikes J; and parameter W; acted as the weight of the To demonstrate LAM defined astrocyte plasticity, we implemented a SNAN with a single-layer of neurons connected to an astrocyte endowed with the BHP learning mechanism (Fig. 7A). To approxi- mate the BHP rule on Loihi, Equation (5) was decomposed into a weighted sum as follows,
Input N; was represented as the integrated current of real-time presynaptic spikes J; and parameter W; acted as the weight of the To demonstrate LAM defined astrocyte plasticity, we implemented a SNAN with a single-layer of neurons connected to an astrocyte endowed with the BHP learning mechanism (Fig. 7A). To approxi- mate the BHP rule on Loihi, Equation (5) was decomposed into a weighted sum as follows,
Related topics:
NeuroscienceComputer ScienceBiologyNeuromorphic Engineering
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved