Browse › Computer Vision › Image Classification › MNIST
MNIST Benchmark (Image Classification)
Image Classification is a fundamental task in vision recognition that aims to understand and categorize an image as a whole under a specific label. Unlike object detection, which involves classification and location of multiple objects within an image, image classification typically pertains to single-object images. When the classification becomes highly detailed or reaches instance-level, it is often referred to as image retrieval, which also involves finding similar images in a large database.
The archive carries no text for this table; the description above is the archive's text for the task Image Classification. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Direction inferred from the metric name, not from the archive: Percentage error (lower is better), Accuracy (higher is better), Cross Entropy Loss (lower is better), Top 1 Accuracy (higher is better). Not inferred (points only, no best-so-far line): Trainable Parameters, Epochs. Points are placed at the row's paper date; 79 of 81 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Branching/Merging CNN + Homogeneous Vector Capsules | 0.13 | 99.87 | 1514187 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 2 | EnsNet (Ensemble learning in CNN augmented with fully connected subnetworks) | 0.16 | 99.84 | – | Paper | Code | 2020 | linked, not harvested | report | ||||
| 3 | Efficient-CapsNet | 0.16 | 99.84 | 161824 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 4 | SOPCNN (Only a single Model) | 0.17 | 99.83 | 1400000 | – | Paper | Code | 2020 | linked, not harvested | report | |||
| 5 | RMDL (30 RDLs) | 0.18 | 99.82 | – | Paper | Code | 2018 | linked, not harvested | report | ||||
| 6 | R-ExplaiNet-22 (single model) | 0.20 | 99.80 | 743882 | – | Paper | Code | 2024 | linked, not harvested | report | |||
| 7 | DropConnect | 0.21 | 99.77 | – | Paper | Code | 2013 | linked, not harvested | report | ||||
| 8 | MCDNN | 0.23 | – | Paper | Code | 2012 | linked, not harvested | report | |||||
| 9 | APAC | 0.23 | – | Paper | – | 2015 | no code linked | report | |||||
| 10 | BNM NiN | 0.24 | – | Paper | Code | 2015 | 3 of 3 ran · 0 unverified | report | |||||
| 11 | SimpleNetv1 | 0.25 | – | Paper | Code | 2016 | 3 of 11 ran · 8 unverified | report | |||||
| 12 | CapsNet | 0.25 | – | Paper | Code | 2017 | 18 of 118 ran · 100 unverified | report | |||||
| 13 | VGG8B + LocalLearning + CO | 0.26 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 14 | VGG-5 (Spinal FC) | 0.28 | 99.72 | – | Paper | Code | 2020 | linked, not harvested | report | ||||
| 15 | TextCaps | 0.29 | 99.71 | – | Paper | Code | 2019 | linked, not harvested | report | ||||
| 16 | ExquisiteNetV2 | 0.29 | 99.71 | 518230 | – | Paper | Code | 2021 | linked, not harvested | report | |||
| 17 | WaveMix-128/7 | 0.29 | – | Paper | Code | 2022 | 1 of 5 ran · 4 unverified | report | |||||
| 18 | Fractional MP | 0.3 | – | Paper | Code | 2014 | 0 of 3 ran · 3 unverified | report | |||||
| 19 | Tree+Max-Avg pooling | 0.3 | – | Paper | Code | 2015 | 1 of 12 ran · 11 unverified | report | |||||
| 20 | CMsC | 0.3 | – | Paper | – | 2015 | no code linked | report | |||||
| 21 | EXACT (M3-CNN) | 0.33 | – | Paper | Code | 2022 | linked, not harvested | report | |||||
| 22 | Second Order Neural Ordinary Differential Equation | 0.37 | 99.63 | – | Paper | Code | 2020 | 1 of 2 ran · 1 unverified | report | ||||
| 23 | Augmented Neural Ordinary Differential Equation | 0.37 | 99.63 | – | Paper | Code | 2019 | 2 of 9 ran · 7 unverified | report | ||||
| 24 | DSN | 0.4 | – | Paper | Code | 2014 | 0 of 21 ran · 21 unverified | report | |||||
| 25 | CKN | 0.4 | – | Paper | – | 2014 | no code linked | report | |||||
| 26 | C-SVDDNet | 0.4 | – | Paper | – | 2014 | no code linked | report | |||||
| 27 | HOPE | 0.4 | – | Paper | – | 2015 | no code linked | report | |||||
| 28 | FLSCNN | 0.4 | – | Paper | – | 2015 | no code linked | report | |||||
| 29 | MIM | 0.4 | – | Paper | – | 2015 | no code linked | report | |||||
| 30 | Fitnet-LSUV-SVM | 0.4 | – | Paper | Code | 2015 | 0 of 19 ran · 19 unverified | report | |||||
| 31 | TAAF-CNN | 0.48% | 99.52% | 421642 | 0.0188 | 35 | – | Paper | Code | 2025 | linked, not harvested | report | |
| 32 | Neural Architecture Search (NAS)-enabled Convolutional Neural Network (CNN) | 0.5 | 99.5 | 1882602 | – | – | – | not matched | report | ||||
| 33 | Maxout Networks | 0.5 | – | Paper | Code | 2013 | 0 of 3 ran · 3 unverified | report | |||||
| 34 | NiN | 0.5 | – | Paper | Code | 2013 | 1 of 7 ran · 6 unverified | report | |||||
| 35 | ReNet | 0.5 | – | Paper | Code | 2015 | linked, not harvested | report | |||||
| 36 | DCNN+GFE | 0.5 | – | Paper | – | 2017 | no code linked | report | |||||
| 37 | VDN | 0.5 | – | Paper | Code | 2015 | linked, not harvested | report | |||||
| 38 | NeuPDE | 0.51 | – | Paper | – | 2019 | no code linked | report | |||||
| 39 | Simple CNN with BaikalCMA loss | 0.53 | – | Paper | Code | 2019 | linked, not harvested | report | |||||
| 40 | SEER (RegNet10B) | 0.58 | 99.42 | – | Paper | Code | 2022 | linked, not harvested | report | ||||
| 41 | Convolutional Tsetlin Machine | 0.6 | 99.4 | – | Paper | Code | 2019 | 0 of 1 ran · 1 unverified | report | ||||
| 42 | PCANet | 0.6 | – | Paper | Code | 2014 | 0 of 3 ran · 3 unverified | report | |||||
| 43 | DiffPrune (LeNet5) | 0.6 | – | Paper | Code | 2020 | linked, not harvested | report | |||||
| 44 | Deep Fried Convnets | 0.7 | – | Paper | Code | 2014 | linked, not harvested | report | |||||
| 45 | Sparse Activity and Sparse Connectivity in Supervised Learning | 0.8 | – | Paper | – | 2016 | no code linked | report | |||||
| 46 | Explaining and Harnessing Adversarial Examples | 0.8 | – | Paper | Code | 2014 | 10 of 21 ran · 11 unverified | report | |||||
| 47 | BinaryConnect | 1.0 | – | Paper | Code | 2015 | 3 of 3 ran · 0 unverified | report | |||||
| 48 | Convolutional PMM (Parametric Matrix Model) | 1.01 | 98.99 | 129416 | – | Paper | – | 2024 | no code linked | report | |||
| 49 | LeNet 300-100 (Sparse Momentum) | 1.26 | – | Paper | Code | 2019 | 0 of 1 ran · 1 unverified | report | |||||
| 50 | Convolutional Clustering | 1.4 | – | Paper | – | 2015 | no code linked | report | |||||
| 51 | CNN Model by Som | 1.41 | 98.59 | – | Paper | Code | 2017 | linked, not harvested | report | ||||
| 52 | Weighted Tsetlin Machine | 1.5 | 98.5 | – | Paper | Code | 2019 | linked, not harvested | report | ||||
| 53 | MLP (ideal number of groups) | 1.67 | – | Paper | – | 2023 | no code linked | report | |||||
| 54 | Perceptron with a tensor train layer | 1.8 | 98.2 | – | Paper | Code | 2015 | 1 of 1 ran · 0 unverified | report | ||||
| 55 | ANODE | 1.8 | 98.2 | – | Paper | Code | 2019 | 2 of 9 ran · 7 unverified | report | ||||
| 56 | Tsetlin Machine | 1.8 | 98.2 | – | Paper | Code | 2018 | 1 of 1 ran · 0 unverified | report | ||||
| 57 | GECCO | 1.96 | 98.04 | – | Paper | Code | 2024 | linked, not harvested | report | ||||
| 58 | PMM (Parametric Matrix Model) | 2.62 | 97.38 | 4990 | – | Paper | – | 2024 | no code linked | report | |||
| 59 | DNN-5 (Trainable Activations) | 2.8 | 97.2 | 575051 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 60 | DNN-3 (Trainable Activations) | 3.0 | 97.0 | 386719 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 61 | DNN-2 (Trainable Activations) | 3.6 | 96.4 | 311651 | – | Paper | Code | 2023 | linked, not harvested | report | |||
| 62 | Zhao et al. (2015) (auto-encoder) | 4.76 | – | Paper | Code | 2015 | linked, not harvested | report | |||||
| 63 | ProjectionNet | 5.0 | 95.0 | – | Paper | – | 2017 | no code linked | report | ||||
| 64 | µ2Net (ViT-L/16) | 99.75 | – | Paper | Code | 2022 | linked, not harvested | report | |||||
| 65 | MobileNet_XnODR | 99.68 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 66 | ResNet-9 | 99.68 | – | Paper | Code | 2022 | linked, not harvested | report | |||||
| 67 | LR-Net | 99.47 | – | Paper | Code | 2022 | linked, not harvested | report | |||||
| 68 | CNN+ Wilson-Cowan model RNN | 99.31 | – | Paper | Code | 2024 | linked, not harvested | report | |||||
| 69 | FastSNN (CNN) | 99.3 | – | Paper | Code | 2022 | 0 of 2 ran · 2 unverified | report | |||||
| 70 | rKAN | 99.293 | – | Paper | Code | 2024 | 2 of 2 ran · 0 unverified | report | |||||
| 71 | CNN-5 Layer | 99.27 | – | Paper | Code | 2021 | 0 of 3 ran · 3 unverified | report | |||||
| 72 | fKAN | 99.228 | – | Paper | Code | 2024 | 1 of 1 ran · 0 unverified | report | |||||
| 73 | StiDi-BP in R-CSNN | 99.2 | – | Paper | – | 2021 | no code linked | report | |||||
| 74 | Wilson-Cowan model RNN | 98.13 | – | Paper | Code | 2024 | linked, not harvested | report | |||||
| 75 | Hypervector Tsetlin Machine | 98.13 | – | Paper | – | 2024 | no code linked | report | |||||
| 76 | ViT-Mini_D9 | 98.03 | 1208586 | – | – | – | not matched | report | |||||
| 77 | FastSNN (MLP) | 97.91 | – | Paper | Code | 2022 | 0 of 2 ran · 2 unverified | report | |||||
| 78 | Binarized MLP with on-chip spiking backpropagation (on Loihi) | 96.2 | – | Paper | Code | 2021 | linked, not harvested | report | |||||
| 79 | SNNL-5 | 95.5 | – | Paper | Code | 2020 | linked, not harvested | report | |||||
| 80 | pFedBreD_ns_mg | 92.47 | – | Paper | – | 2022 | no code linked | report | |||||
| 81 | DGMMC-S | 70 | – | Paper | Code | 2024 | linked, not harvested | report |
All 81 rows shown. 79 link to a paper page on this site; 0 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 25 rows have a graph line, from 23 distinct papers; 15 rows (14 papers) have at least one sample that ran. Counting each paper once: Syntology ran 48 of 252 samples; 204 unverified. Separately, 33 of those 252 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections