A Probabilistic Synapse With Strained MTJs for Spiking Neural Networks
Summary
This study introduces a novel stochastic spiking neural network (SNN) using magnetic tunnel junction (MTJ) synaptic units for efficient, parallel processing. The SNN demonstrates high classification accuracy on handwritten digits, overcoming conventional computing limitations.
Area of Science:
- Neuromorphic Engineering
- Computer Science
- Materials Science
Background:
- Conventional computing faces a significant memory-processor bottleneck.
- Spiking neural networks (SNNs) offer a promising alternative for efficient computation.
- Specialized hardware is needed to realize the full potential of SNNs.
Purpose of the Study:
- To present a stochastic SNN architecture utilizing novel logic-in-memory synaptic units.
- To demonstrate a processing system with massively parallel processing power.
- To evaluate the performance and efficiency of the proposed SNN architecture.
Main Methods:
- Developed a synaptic unit combining strained magnetic tunnel junction (MTJ) devices and transistors.
- Implemented neurons as integrate-and-fire components with thresholding and refraction.
- Fabricated the circuit using CMOS 28-nm technology compatible with MTJ technology.
Main Results:
- The proposed synapse requires minimal area ([Formula: see text]) and low power consumption (675 pW idle, 8.87 fJ per spike).
- The SNN successfully learned and classified handwritten digits from the MNIST database without supervision.
- The network exhibited high classification efficiency despite fabrication variability.
Conclusions:
- The proposed stochastic SNN architecture with MTJ-based synaptic units offers a powerful solution for overcoming the memory-processor bottleneck.
- This design enables highly parallel and energy-efficient neuromorphic computing.
- The demonstrated performance highlights the potential of this approach for advanced AI applications.
Related Concept Videos
02:47The Synapse
132.9K
Neurons communicate with one another by passing on their electrical signals to other neurons. A synapse is the location where two neurons meet to exchange signals. At the synapse, the neuron that sends the signal is called the presynaptic cell, while the neuron that receives the message is called the postsynaptic cell. Note that most neurons can be both presynaptic and postsynaptic, as they both transmit and receive information.
132.9K
02:26Protein Networks
4.5K
An organism can have thousands of different proteins, and these proteins must cooperate to ensure the health of an organism. Proteins bind to other proteins and form complexes to carry out their functions. Many proteins interact with multiple other proteins creating a complex network of protein interactions.
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
4.5K
01:26Chemical Synapses
11.4K
Chemical synapses are specialized sites between two neurons or between a neuron and a non-neuronal cell like a muscle, glandular or sensory cell.
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
11.4K
01:26Chemical Synapses
4.4K
Chemical synapses are specialized sites between two neurons or between a neuron and a non-neuronal cell like a muscle, glandular or sensory cell.
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
4.4K
01:28Electrical Synapses
10.3K
Electrical synapses found in all nervous systems play important and unique roles. In these synapses, the presynaptic and postsynaptic membranes are very close together (3.5 nm) and are actually physically connected by channel proteins forming gap junctions.
Gap junctions allow the current to pass directly from one cell to the next. In contrast, in the chemical synapse, the neurotransmitters carry the information through the synaptic cleft from one neuron to the next. They consist of two...
Gap junctions allow the current to pass directly from one cell to the next. In contrast, in the chemical synapse, the neurotransmitters carry the information through the synaptic cleft from one neuron to the next. They consist of two...
10.3K
02:18Network Covalent Solids
16.1K
Network covalent solids contain a three-dimensional network of covalently bonded atoms as found in the crystal structures of nonmetals like diamond, graphite, silicon, and some covalent compounds, such as silicon dioxide (sand) and silicon carbide (carborundum, the abrasive on sandpaper). Many minerals have networks of covalent bonds.
To break or to melt a covalent network solid, covalent bonds must be broken. Because covalent bonds are relatively strong, covalent network solids are typically...
To break or to melt a covalent network solid, covalent bonds must be broken. Because covalent bonds are relatively strong, covalent network solids are typically...
16.1K