Hugging Face's logo

Datasets:
FL-Security
/
Poisoning_Resilient_Federated_Learning_Playground

Tasks:
Tabular Classification
Modalities:
Text
Formats:
text
Size:
1M - 10M
Tags:
federated-learning
poisoning-attacks
robust-aggregation
benchmark
experiment-results
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
text
stringlengths
0
1.35k
seed: 42, num_experiments: 5, experiment_id: 0, epochs: 200, algorithm: FedAvg, optimizer: SGD, momentum: 0.9, weight_decay: 0.0005, lr_scheduler: MultiStepLR, milestones: [0.5, 0.8], num_clients: 20, batch_size: 64, learning_rate: 0.05, local_epochs: 4, model: resnet18, dataset: CIFAR100, distribution: iid, im_iid_gam...
Started on Thu Feb 26 03:22:13 2026
Generating new indices
Doing iid partition
iid partition finished
Data partitioned
Clients and server are initialized
Starting Training...
Epoch 0 Train Acc: 0.0135 Train loss: 0.0750 Test Acc: 0.0099 Test loss: 0.0723
Epoch 1 Train Acc: 0.0189 Train loss: 0.0741 Test Acc: 0.0107 Test loss: 0.0725
Epoch 2 Train Acc: 0.0311 Train loss: 0.0721 Test Acc: 0.0126 Test loss: 0.0726
Epoch 3 Train Acc: 0.0375 Train loss: 0.0705 Test Acc: 0.0285 Test loss: 0.0703
Epoch 4 Train Acc: 0.0465 Train loss: 0.0691 Test Acc: 0.0633 Test loss: 0.0665
Epoch 5 Train Acc: 0.0480 Train loss: 0.0676 Test Acc: 0.0634 Test loss: 0.0643
Epoch 6 Train Acc: 0.0588 Train loss: 0.0663 Test Acc: 0.0600 Test loss: 0.0675
Epoch 7 Train Acc: 0.0637 Train loss: 0.0648 Test Acc: 0.0625 Test loss: 0.0662
Epoch 8 Train Acc: 0.0695 Train loss: 0.0640 Test Acc: 0.0735 Test loss: 0.0644
Epoch 9 Train Acc: 0.0596 Train loss: 0.3417 Test Acc: 0.0776 Test loss: 0.0639
Epoch 10 Train Acc: 0.0441 Train loss: 0.0732 Test Acc: 0.0340 Test loss: 0.1593
Epoch 11 Train Acc: 0.0537 Train loss: 0.0661 Test Acc: 0.0399 Test loss: 0.1554
Epoch 12 Train Acc: 0.0619 Train loss: 0.0647 Test Acc: 0.0617 Test loss: 0.0777
Epoch 13 Train Acc: 0.0762 Train loss: 0.0630 Test Acc: 0.0823 Test loss: 0.0665
Epoch 14 Train Acc: 0.0867 Train loss: 0.0613 Test Acc: 0.1033 Test loss: 0.0613
Epoch 15 Train Acc: 0.0887 Train loss: 0.0607 Test Acc: 0.1162 Test loss: 0.0589
Epoch 16 Train Acc: 0.0910 Train loss: 0.0609 Test Acc: 0.1226 Test loss: 0.0583
Epoch 17 Train Acc: 0.0930 Train loss: 0.0602 Test Acc: 0.1246 Test loss: 0.0581
Epoch 18 Train Acc: 0.0943 Train loss: 0.0603 Test Acc: 0.1271 Test loss: 0.0576
Epoch 19 Train Acc: 0.0916 Train loss: 0.3111 Test Acc: 0.1316 Test loss: 0.0581
Epoch 20 Train Acc: 0.0615 Train loss: 0.0674 Test Acc: 0.1230 Test loss: 0.0585
Epoch 21 Train Acc: 0.1004 Train loss: 0.0604 Test Acc: 0.1149 Test loss: 0.0599
Epoch 22 Train Acc: 0.1090 Train loss: 0.0594 Test Acc: 0.1365 Test loss: 0.0574
Epoch 23 Train Acc: 0.1086 Train loss: 0.0590 Test Acc: 0.1411 Test loss: 0.0566
Epoch 24 Train Acc: 0.1225 Train loss: 0.0581 Test Acc: 0.1438 Test loss: 0.0564
Epoch 25 Train Acc: 0.1229 Train loss: 0.0576 Test Acc: 0.1499 Test loss: 0.0558
Epoch 26 Train Acc: 0.1154 Train loss: 0.0578 Test Acc: 0.1563 Test loss: 0.0556
Epoch 27 Train Acc: 0.1348 Train loss: 0.0568 Test Acc: 0.1647 Test loss: 0.0547
Epoch 28 Train Acc: 0.1414 Train loss: 0.0565 Test Acc: 0.1667 Test loss: 0.0546
Epoch 29 Train Acc: 0.1242 Train loss: 0.3045 Test Acc: 0.1616 Test loss: 0.0548
Epoch 30 Train Acc: 0.0771 Train loss: 0.0651 Test Acc: 0.1358 Test loss: 0.0576
Epoch 31 Train Acc: 0.1187 Train loss: 0.0583 Test Acc: 0.1484 Test loss: 0.0563
Epoch 32 Train Acc: 0.1412 Train loss: 0.0568 Test Acc: 0.1517 Test loss: 0.0555
Epoch 33 Train Acc: 0.1461 Train loss: 0.0565 Test Acc: 0.1625 Test loss: 0.0545
Epoch 34 Train Acc: 0.1502 Train loss: 0.0555 Test Acc: 0.1743 Test loss: 0.0539
Epoch 35 Train Acc: 0.1553 Train loss: 0.0550 Test Acc: 0.1831 Test loss: 0.0535
Epoch 36 Train Acc: 0.1445 Train loss: 0.0551 Test Acc: 0.1944 Test loss: 0.0526
Epoch 37 Train Acc: 0.1602 Train loss: 0.0540 Test Acc: 0.1910 Test loss: 0.0522
Epoch 38 Train Acc: 0.1594 Train loss: 0.0542 Test Acc: 0.2006 Test loss: 0.0515
Epoch 39 Train Acc: 0.1568 Train loss: 0.2813 Test Acc: 0.1904 Test loss: 0.0520
Epoch 40 Train Acc: 0.1102 Train loss: 0.0620 Test Acc: 0.1846 Test loss: 0.0527
Epoch 41 Train Acc: 0.1596 Train loss: 0.0554 Test Acc: 0.1784 Test loss: 0.0538
Epoch 42 Train Acc: 0.1691 Train loss: 0.0536 Test Acc: 0.1961 Test loss: 0.0519
Epoch 43 Train Acc: 0.1760 Train loss: 0.0533 Test Acc: 0.2109 Test loss: 0.0509
Epoch 44 Train Acc: 0.1758 Train loss: 0.0528 Test Acc: 0.2201 Test loss: 0.0498
Epoch 45 Train Acc: 0.1842 Train loss: 0.0522 Test Acc: 0.2256 Test loss: 0.0498
Epoch 46 Train Acc: 0.1805 Train loss: 0.0524 Test Acc: 0.2244 Test loss: 0.0497
Epoch 47 Train Acc: 0.1945 Train loss: 0.0512 Test Acc: 0.2282 Test loss: 0.0495
Epoch 48 Train Acc: 0.1973 Train loss: 0.0516 Test Acc: 0.2306 Test loss: 0.0489
Epoch 49 Train Acc: 0.1822 Train loss: 0.2909 Test Acc: 0.2144 Test loss: 0.0501
Epoch 50 Train Acc: 0.1242 Train loss: 0.0601 Test Acc: 0.1968 Test loss: 0.0513
Epoch 51 Train Acc: 0.1771 Train loss: 0.0534 Test Acc: 0.2172 Test loss: 0.0519
Epoch 52 Train Acc: 0.1730 Train loss: 0.0524 Test Acc: 0.2134 Test loss: 0.0510
Epoch 53 Train Acc: 0.1982 Train loss: 0.0515 Test Acc: 0.2249 Test loss: 0.0495
Epoch 54 Train Acc: 0.2074 Train loss: 0.0506 Test Acc: 0.2441 Test loss: 0.0482
Epoch 55 Train Acc: 0.2213 Train loss: 0.0499 Test Acc: 0.2484 Test loss: 0.0477
Epoch 56 Train Acc: 0.2092 Train loss: 0.0502 Test Acc: 0.2445 Test loss: 0.0476
Epoch 57 Train Acc: 0.2244 Train loss: 0.0487 Test Acc: 0.2511 Test loss: 0.0473
Epoch 58 Train Acc: 0.2252 Train loss: 0.0488 Test Acc: 0.2593 Test loss: 0.0469
Epoch 59 Train Acc: 0.2119 Train loss: 0.2637 Test Acc: 0.2519 Test loss: 0.0469
Epoch 60 Train Acc: 0.1617 Train loss: 0.0564 Test Acc: 0.2383 Test loss: 0.0489
Epoch 61 Train Acc: 0.2113 Train loss: 0.0506 Test Acc: 0.2399 Test loss: 0.0483
Epoch 62 Train Acc: 0.2229 Train loss: 0.0490 Test Acc: 0.2502 Test loss: 0.0473
Epoch 63 Train Acc: 0.2234 Train loss: 0.0486 Test Acc: 0.2619 Test loss: 0.0463
Epoch 64 Train Acc: 0.2393 Train loss: 0.0481 Test Acc: 0.2530 Test loss: 0.0470
Epoch 65 Train Acc: 0.2271 Train loss: 0.0484 Test Acc: 0.2634 Test loss: 0.0459
Epoch 66 Train Acc: 0.2318 Train loss: 0.0476 Test Acc: 0.2802 Test loss: 0.0446
Epoch 67 Train Acc: 0.2469 Train loss: 0.0470 Test Acc: 0.2837 Test loss: 0.0443
Epoch 68 Train Acc: 0.2437 Train loss: 0.0470 Test Acc: 0.2922 Test loss: 0.0440
Epoch 69 Train Acc: 0.2510 Train loss: 0.2598 Test Acc: 0.2670 Test loss: 0.0456
Epoch 70 Train Acc: 0.1691 Train loss: 0.0553 Test Acc: 0.2468 Test loss: 0.0481
FLDetector Defense: Benign idx: [0 1]
Epoch 71 Train Acc: 0.2348 Train loss: 0.0487 Test Acc: 0.2512 Test loss: 0.0476
FLDetector Defense: Benign idx: [0 1]
Epoch 72 Train Acc: 0.2471 Train loss: 0.0468 Test Acc: 0.2798 Test loss: 0.0451
FLDetector Defense: Benign idx: [0 1]
Epoch 73 Train Acc: 0.2414 Train loss: 0.0467 Test Acc: 0.2972 Test loss: 0.0438
FLDetector Defense: Benign idx: [0 1]
Epoch 74 Train Acc: 0.2469 Train loss: 0.0463 Test Acc: 0.2987 Test loss: 0.0433
FLDetector Defense: Benign idx: [0 1]
Epoch 75 Train Acc: 0.2652 Train loss: 0.0453 Test Acc: 0.2874 Test loss: 0.0435
FLDetector Defense: Benign idx: [ 0 1 14 16]
Epoch 76 Train Acc: 0.2664 Train loss: 0.0450 Test Acc: 0.2751 Test loss: 0.0445
FLDetector Defense: Benign idx: [ 0 1 14 16]
Epoch 77 Train Acc: 0.2596 Train loss: 0.0458 Test Acc: 0.2788 Test loss: 0.0450
FLDetector Defense: Benign idx: [ 0 1 14]
Epoch 78 Train Acc: 0.2574 Train loss: 0.0462 Test Acc: 0.2674 Test loss: 0.0462
FLDetector Defense: Benign idx: [0 1]
Epoch 79 Train Acc: 0.2506 Train loss: 0.2546 Test Acc: 0.2849 Test loss: 0.0443
FLDetector Defense: Benign idx: [0 1]
Epoch 80 Train Acc: 0.1758 Train loss: 0.0549 Test Acc: 0.2649 Test loss: 0.0467
End of preview. Expand in Data Studio

FL Security Experiment Results

This repository contains the experiment outputs used for the FL Security / FLPoison federated learning poisoning benchmark. The archive is intended for readers who want to inspect the raw training logs, reuse the aggregated curves and tables, or reproduce the paper figures without rerunning the full Compute Canada workload.

The uploaded artifact is:

exp_data.tar.gz  # about 241 MB

After extraction, the archive keeps the original Compute Canada scratch layout:

scratch/FL_Poison/
  logs_from_slurm_split_20260312/
  replot_from_slurm_no300_noadv04_20260312/
  tables_test_loss_from_slurm_split_cleaned_merged_ranked_epochmerged_nomean_withbase/
  custom_compare_fangattack_mean_vs_noattack/

The FL_Poison directory name is a legacy path from the experiment runs. The code repository has since been renamed to FL_Security.

What Is Included

  • logs_from_slurm_split_20260312/: split raw result logs recovered from Slurm outputs.
  • replot_from_slurm_no300_noadv04_20260312/: regenerated plots and plot indexes after filtering known incomplete or out-of-scope runs.
  • tables_test_loss_from_slurm_split_cleaned_merged_ranked_epochmerged_nomean_withbase/: LaTeX tables and an index for final test-loss comparisons.
  • custom_compare_fangattack_mean_vs_noattack/: focused FangAttack vs. NoAttack comparisons for selected settings.

The archive contains result files only. It does not include the original training datasets such as CIFAR10, CIFAR100, or TinyImageNet.

Experiment Scope

The main result set covers federated learning poisoning experiments with:

  • Algorithms: FedAvg
  • Datasets/models: CIFAR10 + VGG19, CIFAR100 + ResNet18
  • Data splits: iid, non-iid
  • Non-IID Dirichlet alpha values: 0.1, 0.5, 1
  • Adversarial client fractions: 0.1, 0.2, 0.3, with some raw logs also containing 0.4
  • Seeds used in the regenerated plots: 42, 43, 44, 45, 46
  • Attacks in the main plots: ALIE, FangAttack, MinMax, MinSum

Most regenerated figures target the 200-round configuration and use epoch 199 / 200 as the final evaluation point, depending on whether the file is indexed by zero-based epoch or by configured training rounds.

Important Files

The most useful entry points are:

scratch/FL_Poison/logs_from_slurm_split_20260312/extraction_summary.txt
scratch/FL_Poison/logs_from_slurm_split_20260312/extraction_manifest.tsv

scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/accuracy_1x4/summary.txt
scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/accuracy_1x4/plot_index.csv

scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/adv_sensitivity_0.1-0.3_epoch199/plot_index.csv
scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/alpha_non_iid_epoch199/plot_index.csv
scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/num_clients_sensitivity_epoch199/figure_index.csv
scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/triguard_reputation_curves/plot_index.csv
scratch/FL_Poison/replot_from_slurm_no300_noadv04_20260312/triguard_fp_fn/summary.txt

scratch/FL_Poison/tables_test_loss_from_slurm_split_cleaned_merged_ranked_epochmerged_nomean_withbase/table_index.csv
scratch/FL_Poison/tables_test_loss_from_slurm_split_cleaned_merged_ranked_epochmerged_nomean_withbase/summary.txt

The index CSV files are the easiest way to locate a figure or table by dataset, model, IID mode, adversarial fraction, alpha, client count, and seed coverage.

Quick Start

Download and extract the archive:

tar -xzf exp_data.tar.gz
cd scratch/FL_Poison

Inspect the raw extraction summary:

sed -n '1,80p' logs_from_slurm_split_20260312/extraction_summary.txt

List available regenerated plots:

python - <<'PY'
import pandas as pd

idx = pd.read_csv(
    "replot_from_slurm_no300_noadv04_20260312/accuracy_1x4/plot_index.csv"
)
print(idx[["iid", "dataset", "model", "num_clients", "adv", "alpha", "seeds_used", "figure_png"]].head())
PY

Open a generated plot directly from the figure_png or figure_pdf path recorded in the index. If you extracted the archive outside /home/fengye, replace the absolute prefix in the index paths with your local extraction path.

Directory Details

Raw Slurm-Split Logs

logs_from_slurm_split_20260312/ contains split result logs extracted from Compute Canada Slurm output files. The summary reports:

out_files=2140
total_seed_segments=10089
result_txt_total=9622
dataset_seen={'CIFAR10': 4351, 'CIFAR100': 5738}
iid_seen={'iid': 2521, 'non-iid': 7568}
adv_seen={'0.1': 2383, '0.2': 3105, '0.3': 2831, '0.4': 1770}

Use extraction_manifest.tsv to map extracted text files back to their Slurm source segments.

Regenerated Figures

replot_from_slurm_no300_noadv04_20260312/ contains PNG/PDF figures and CSV indexes. The main subdirectories are:

  • accuracy_1x4/: attack/defense accuracy curves grouped into 1x4 panels.
  • adv_sensitivity_0.1-0.3_epoch199/: sensitivity to adversarial client fraction.
  • alpha_non_iid_epoch199/: sensitivity to Dirichlet non-IID alpha.
  • num_clients_sensitivity_epoch199/: sensitivity to number of clients.
  • triguard_reputation_curves/: TriGuardFL reputation curves.
  • triguard_fp_fn/: TriGuardFL false-positive/false-negative summaries.
  • all_figures/: a flat collection of generated figure files.

The accuracy_1x4/summary.txt file records the filtering used for the main regenerated plots:

parsed_after_seed_filter=6901
filtered_by_adv=1598
filtered_by_epoch_cfg=1123
groups_total=48
generated_figures=48
max_epoch=200
attacks=ALIE,FangAttack,MinMax,MinSum
seed_filter=42,43,44,45,46
adv_filter=0.1,0.2,0.3
exclude_config_epochs=300
filter_buggy_fltrust=True

Tables

tables_test_loss_from_slurm_split_cleaned_merged_ranked_epochmerged_nomean_withbase/ contains LaTeX tables for final test-loss comparisons. table_index.csv gives the table path and metadata fields:

table_path,iid,dataset,model,num_clients,lr,algo,adv,alpha,cfg,epochs_tag,loss_stat

Tables are grouped by dataset and adversarial fraction, for example:

CIFAR100/adv_0.2/
CIFAR100/adv_0.3/
CIFAR100/adv_0.4/
CIFAR10/adv_0.2/
CIFAR10/adv_0.3/
CIFAR10/adv_0.4/

Reproducing or Extending the Analysis

The code used to run and post-process experiments is available in the GitHub repository:

https://github.com/fye97/FL_Security

Relevant local entry points in that repository include:

exps/run_cc.sh
exps/run_cc_carat.sh
exps/launch.py
plot_num_clients_accuracy_sensitivity.py

For new Compute Canada runs, use the experiment launcher rather than submitting worker scripts directly:

./exps/run_cc.sh <spec> --dry-run
./exps/run_cc.sh <spec>

Notes and Caveats

  • Some raw logs include adversarial fraction 0.4, but the main regenerated figures filter to 0.1, 0.2, and 0.3.
  • Some 300-round configurations were excluded from the regenerated figure set to keep the plotted results aligned with the 200-round analysis.
  • The result indexes may contain absolute paths from the original machine. Treat those paths as metadata; replace the /home/fengye/scratch/FL_Poison prefix with your local extracted path when opening files programmatically.
  • The raw text logs are useful for auditability, but most readers should start from the plot_index.csv, figure_index.csv, and table_index.csv files.
Downloads last month
333
Number of rows:
5,738,575
Total file size:
558 MB