{ "schema": "pulse-split/1", "name": "pool33", "data_tag": "v1.1", "used_by": [ "T1", "T4", "T7" ], "n_volunteers": 33, "volunteers": [ "v1", "v2", "v3", "v9", "v11", "v12", "v13", "v15", "v16", "v17", "v19", "v20", "v21", "v22", "v23", "v24", "v25", "v26", "v27", "v28", "v29", "v30", "v31", "v32", "v33", "v34", "v35", "v36", "v37", "v38", "v39", "v40", "v41" ], "n_folds": 5, "folds": [ { "fold": 0, "test": [ "v1", "v12", "v19", "v24", "v29", "v34", "v39" ], "inner_val": [ "v16", "v25", "v33", "v36" ], "train": [ "v2", "v3", "v9", "v11", "v13", "v15", "v17", "v20", "v21", "v22", "v23", "v26", "v27", "v28", "v30", "v31", "v32", "v35", "v37", "v38", "v40", "v41" ], "n": { "test": 7, "inner_val": 4, "train": 22 }, "inner_val_strata": { "none": [ "v16" ], "pressure": [ "v25", "v33", "v36" ] }, "train_strata_sizes": { "none": 6, "pressure": 20 } }, { "fold": 1, "test": [ "v2", "v13", "v20", "v25", "v30", "v35", "v40" ], "inner_val": [ "v15", "v26", "v29", "v41" ], "train": [ "v1", "v3", "v9", "v11", "v12", "v16", "v17", "v19", "v21", "v22", "v23", "v24", "v27", "v28", "v31", "v32", "v33", "v34", "v36", "v37", "v38", "v39" ], "n": { "test": 7, "inner_val": 4, "train": 22 }, "inner_val_strata": { "none": [ "v15" ], "pressure": [ "v26", "v29", "v41" ] }, "train_strata_sizes": { "none": 6, "pressure": 20 } }, { "fold": 2, "test": [ "v3", "v15", "v21", "v26", "v31", "v36", "v41" ], "inner_val": [ "v11", "v20", "v27", "v40" ], "train": [ "v1", "v2", "v9", "v12", "v13", "v16", "v17", "v19", "v22", "v23", "v24", "v25", "v28", "v29", "v30", "v32", "v33", "v34", "v35", "v37", "v38", "v39" ], "n": { "test": 7, "inner_val": 4, "train": 22 }, "inner_val_strata": { "none": [ "v20" ], "pressure": [ "v11", "v27", "v40" ] }, "train_strata_sizes": { "none": 5, "pressure": 21 } }, { "fold": 3, "test": [ "v9", "v16", "v22", "v27", "v32", "v37" ], "inner_val": [ "v13", "v21", "v24", "v38" ], "train": [ "v1", "v2", "v3", "v11", "v12", "v15", "v17", "v19", "v20", "v23", "v25", "v26", "v28", "v29", "v30", "v31", "v33", "v34", "v35", "v36", "v39", "v40", "v41" ], "n": { "test": 6, "inner_val": 4, "train": 23 }, "inner_val_strata": { "none": [ "v21" ], "pressure": [ "v13", "v24", "v38" ] }, "train_strata_sizes": { "none": 5, "pressure": 22 } }, { "fold": 4, "test": [ "v11", "v17", "v23", "v28", "v33", "v38" ], "inner_val": [ "v19", "v30", "v34", "v39" ], "train": [ "v1", "v2", "v3", "v9", "v12", "v13", "v15", "v16", "v20", "v21", "v22", "v24", "v25", "v26", "v27", "v29", "v31", "v32", "v35", "v36", "v37", "v40", "v41" ], "n": { "test": 6, "inner_val": 4, "train": 23 }, "inner_val_strata": { "none": [ "v19" ], "pressure": [ "v30", "v34", "v39" ] }, "train_strata_sizes": { "none": 6, "pressure": 21 } } ], "source": { "file": "results/t1_cv/cv_folds.json", "md5": "8f20127a120ad516bedc52c983e82eb2", "note": "test folds copied verbatim (order preserved); the source's data_root field is not reproduced", "fold_sizes": [ 7, 7, 7, 6, 6 ], "generator": "round-robin over the pool sorted by volunteer number: fold k = pool[k::5] (re-derivable; documented in the original CV notes)" }, "inner_val_rule": { "fraction": 0.15, "required_modalities": [ "emg", "mocap" ], "strata_modalities": [ "pressure" ], "require_annotations": false, "metadata": "metadata/recordings.csv (aligned_modalities, has_dense_annotations)", "rule": "1. The test folds are copied verbatim from the source fold file.\n2. Per fold, ``n_val = floor(0.15 * n_train + 0.5)`` of the fold's training\n volunteers become inner-validation volunteers (pool33: 4, pool27: 3).\n3. Strata: volunteers are grouped by whether a majority of their S1-S8\n recordings list each *strata modality* in metadata/recordings.csv (pool33:\n pressure, so the 7 volunteers without pressure are represented). ``n_val``\n is split over strata in proportion to their size among the fold's training\n volunteers (largest remainder; ties by stratum key).\n4. Eligibility: an inner-validation volunteer must have all 8 scenarios S1-S8\n with every *required modality* listed (and dense labels, if required), so\n the inner-validation set covers all scenario classes. Volunteers with\n partial coverage always stay in the training set.\n5. Within a stratum, eligible volunteers are ranked by (number of earlier folds\n in which they were already inner-validation, sha256 of\n ``\"pulse-inner-val//fold/\"``) and the first ones are taken, so\n picks are spread over volunteers and fully reproducible." }, "volunteer_profiles": { "v1": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v2": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v3": { "n_task_recordings": 8, "n_full_recordings": 5, "eligible": false, "stratum": "pressure" }, "v9": { "n_task_recordings": 8, "n_full_recordings": 1, "eligible": false, "stratum": "none" }, "v11": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v12": { "n_task_recordings": 8, "n_full_recordings": 6, "eligible": false, "stratum": "pressure" }, "v13": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v15": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "none" }, "v16": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "none" }, "v17": { "n_task_recordings": 8, "n_full_recordings": 7, "eligible": false, "stratum": "none" }, "v19": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "none" }, "v20": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "none" }, "v21": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "none" }, "v22": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v23": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v24": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v25": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v26": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v27": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v28": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v29": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v30": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v31": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v32": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v33": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v34": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v35": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v36": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v37": { "n_task_recordings": 3, "n_full_recordings": 3, "eligible": false, "stratum": "pressure" }, "v38": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v39": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v40": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" }, "v41": { "n_task_recordings": 8, "n_full_recordings": 8, "eligible": true, "stratum": "pressure" } } }