Research Article
An Area-Efficient Integrate-and-Fire Neuron Circuit with Enhanced Robustness against Synapse Variability in Hardware Neural Network
Table 1
Comparison among the performances of reported neuron circuits and the proposed one in this work.
| References | Synaptic input | Neuron model | Circuit type | Technology node (µm) | Firing frequency | Power consumption (μW) | Special features |
| [18] | Current | Conductance-based | Analog | — | ∼300 Hz | 60 | Neuro-physiological principles | [23] | Current | Integrate-and-fire | Analog | 1.5 | ∼1.2 kHz | 1.5 | Spike frequency adaptation | [26] | Current | Hindmarsh–Rose | Mixed mode | 0.25 | ∼2 kHz | 163.4 | Time-scaling techniques | [27] | Current | Izhikevich | Analog | 0.35 | ∼1 MHz | 40 | Firing pattern and shape | [28] | Current | Integrate-and-fire | Mixed mode | 0.8 | ∼100 Hz | 120 | Spike-based learning | [29] | Voltage | Integrate-and-fire | Mixed mode | — | ∼1 kHz | — | Positive feedback neuron | This work | Current | Leaky integrate-and-fire | Analog | 0.35 | ∼5 kHz | ∼14.7 | Resilience against the synaptic device fluctuations |
|
|