# Image Classification on MNIST (Accuracy and Hardware Efficiency)

Source: [Image Classification on MNIST (Accuracy and Hardware Efficiency)](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-19T13:54:22.915Z

Evaluation of Image Classification performance on the MNIST dataset, reporting standard accuracy alongside detailed hardware performance metrics (Frequency, Latency, Power, Energy) and resource utilization (LUTs, FFs, DSPs, BRAMs).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Image Classification |
| Dataset (including any stated split) | MNIST |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Frequency (MHz)](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=Frequency%20(MHz)>): Higher is better; unit: OTHER.

- [Power (mW)](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=Power%20(mW)>): Lower is better; unit: OTHER.

- [Latency (ms)](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=Latency%20(ms)>): Lower is better; unit: TIME.

- [Energy (mJ)](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=Energy%20(mJ)>): Lower is better; unit: OTHER.

- [LUTs](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=LUTs>): Lower is better; unit: SIZE.

- [FFs](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=FFs>): Lower is better; unit: SIZE.

- [DSPs](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=DSPs>): Lower is better; unit: SIZE.

- [BRAMs](<https://www.sota2.com/research/sota/image-classification-on-mnist-accuracy-and-hardware-efficiency.md?metric=BRAMs>): Lower is better; unit: SIZE.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| \[16\] | 97.21 | Not recorded | 6a0beecedbc836b87d82c6e4 |
| \[17\] | 95.1 | Not recorded | 6a0beecedbc836b87d82c6e4 |
| Spiker-LL | 92 | Not recorded | 6a0beecedbc836b87d82c6e4 |
| \[18\] | 90.19 | Not recorded | 6a0beecedbc836b87d82c6e4 |
| Spiker-LL (8-bits) | 90 | Not recorded | 6a0beecedbc836b87d82c6e4 |

### \[16\]

Accuracy: 97.21. Result ID: 6a0c6b79fbc373592c04530c.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

Recorded evaluation configuration: {"Year":"2025","Learning rule":"Optimized AL","Neuron type":"IF","Weight bw \[bit\]":"8 or 16","Neuron bw \[bit\]":"13 or 21","Platform":"Zynq-7045","Architecture":"784-512-100"}. This record does not assert independent reproduction.

### \[17\]

Accuracy: 95.1. Result ID: 6a0c6b79fbc373592c04530e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

Recorded evaluation configuration: {"Year":"2022","Learning rule":"Triple Brain","Neuron type":"LIF","Weight bw \[bit\]":"16","Neuron bw \[bit\]":"24","Platform":"Zynq-7045","Architecture":"784-256"}. This record does not assert independent reproduction.

### Spiker-LL

Accuracy: 92. Result ID: 6a0c6b79fbc373592c04530a.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

Recorded evaluation configuration: {"Year":"This work","Learning rule":"STSF","Neuron type":"LIF","Weight bw \[bit\]":"16","Neuron bw \[bit\]":"16","Platform":"Pynq Z2","Architecture":"784-200-10"}. This record does not assert independent reproduction.

### \[18\]

Accuracy: 90.19. Result ID: 6a0c6b79fbc373592c04530d.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

Recorded evaluation configuration: {"Year":"2024","Learning rule":"PLR","Neuron type":"LIF","Weight bw \[bit\]":"8","Neuron bw \[bit\]":"8","Platform":"Zynq-7045","Architecture":"784-256"}. This record does not assert independent reproduction.

### Spiker-LL (8-bits)

Accuracy: 90. Result ID: 6a0c6b79fbc373592c04530b.

[Reporting paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0beecedbc836b87d82c6e4.md>)

Recorded evaluation configuration: {"Year":"This work","Learning rule":"STSF","Neuron type":"LIF","Weight bw \[bit\]":"8","Neuron bw \[bit\]":"8","Platform":"Pynq Z2","Architecture":"784-200-10"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.