# Byzantine Attack Detection on MNIST (100-client, 80 honest, 20 Byzantine, test) Source: [Byzantine Attack Detection on MNIST (100-client, 80 honest, 20 Byzantine, test)]() Collection index: [llms.txt]() Data updated: 2026-05-19T07:22:55.152Z Evaluation of Byzantine Attack Detection performance on a specific MNIST test split configured with 100 clients (80 honest, 20 Byzantine). Metrics include performance against specific attack types like Clustered, Sign Flip, and ALIE. 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 | Byzantine Attack Detection | | Dataset (including any stated split) | MNIST 100-client, 80 honest, 20 Byzantine (test) | | Metric | Clustered Detection Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 6 of 6 | | 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 - [Clustered Detection Rate](): Higher is better; unit: PERCENT. - [Sign Flip Detection Rate](): Higher is better; unit: PERCENT. - [ALIE Detection Rate](): Higher is better; unit: PERCENT. - [Sparse Lie Detection Rate](): Higher is better; unit: PERCENT. - [Gaussian Noise Detection Rate](): Higher is better; unit: PERCENT. - [Scale Attack Detection Rate](): Higher is better; unit: PERCENT. - [Average Detection Rate](): Higher is better; unit: PERCENT. ## Recorded results | Method | Clustered Detection Rate | Result date | Report paper | | --- | --- | --- | --- | | Classical | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | | Cascaded Dual-QUBO | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | | MultiSignal ensemble | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | | Classical MultiKrum | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | | Cascaded Dual-QUBO | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | | MultiSignal ensemble | 100 | Not recorded | 6a0bee88dbc836b87d82c388 | ### Classical Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b755. [Reporting paper]() Recorded evaluation configuration: {"clients":"100","honest\_clients":"80","byzantine\_clients":"20","rounds":"5"}. This record does not assert independent reproduction. ### Cascaded Dual-QUBO Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b756. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"clients":"100","honest\_clients":"80","byzantine\_clients":"20","rounds":"5"}. This record does not assert independent reproduction. ### MultiSignal ensemble Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b757. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"clients":"100","honest\_clients":"80","byzantine\_clients":"20","rounds":"5"}. This record does not assert independent reproduction. ### Classical MultiKrum Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b7ca. [Reporting paper]() Recorded evaluation configuration: {"Method variant":"Classical","Total Clients":"100","Honest Clients":"80","Byzantine Clients":"20","Rounds":"5"}. This record does not assert independent reproduction. ### Cascaded Dual-QUBO Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b7cb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method variant":"Cascade","Total Clients":"100","Honest Clients":"80","Byzantine Clients":"20","Rounds":"5"}. This record does not assert independent reproduction. ### MultiSignal ensemble Clustered Detection Rate: 100. Result ID: 6a0c0fb1fbc373592c01b7cc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Method variant":"MultiSignal","Total Clients":"100","Honest Clients":"80","Byzantine Clients":"20","Rounds":"5"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.