How fahad3166/temporal-scmvc-p2 works
Temporal SCMVC is a research implementation for multi-view time-series clustering. Its principal workflow is to select a supported sensor dataset and views, load temporal windows, train a reconstruction-plus-contrastive model with a temporal backbone, and assess clustering quality. The repository also contains local-window training variants and preprocessing-index audit utilities. The graph follows README-described capabilities, but sampled source is incomplete, so uncertain evaluation and audit wiring is omitted.
Main entry points: train_temporal_tcn.py TCN experiment · train_wisdm_temporal_local.py Local-window training · train_temporal_lstm.py LSTM experiment · train_temporal_mlp.py MLP experiment
Components 17
Temporal data
- UCI datasets
- HAR loader
temporal_scmvc/dataloader_har_temporal.py - MHEALTH loader
temporal_scmvc/dataloader_mhealth_temporal.py - PAMAP2 loader
temporal_scmvc/dataloader_pamap_temporal.py - WISDM loader
temporal_scmvc/dataloader_wisdm_temporal.py - Sample selection
temporal_scmvc/sampling.py
Model training
- TCN experiment
temporal_scmvc/train_temporal_tcn.py - LSTM experiment
temporal_scmvc/train_temporal_lstm.py - MLP experiment
temporal_scmvc/train_temporal_mlp.py - Local-window training
temporal_scmvc/train_wisdm_temporal_local.py - TCN model
temporal_scmvc/network_temporal_tcn.py - Temporal model
temporal_scmvc/network_temporal.py - LSTM model
temporal_scmvc/network_temporal_lstm.py - Contrastive objective
temporal_scmvc/loss2.py
Evaluation and audit
- Clustering metrics
temporal_scmvc/metric.py - Index audit
temporal_scmvc/save_indices.py
Other components
- Researcher
Connections 19
- Researcher → TCN experiment: starts experiment
- UCI datasets → HAR loader: provides files
- UCI datasets → MHEALTH loader: provides files
- UCI datasets → PAMAP2 loader: provides files
- UCI datasets → WISDM loader: provides files
- TCN experiment → HAR loader: loads dataset
- TCN experiment → MHEALTH loader: loads dataset
- TCN experiment → PAMAP2 loader: loads dataset
- TCN experiment → WISDM loader: loads dataset
- TCN experiment → TCN model: instantiates
- TCN experiment → Contrastive objective: creates objective
- LSTM experiment → LSTM model: instantiates
- MLP experiment → Temporal model: instantiates
- Local-window training → Temporal model: runs model
- Local-window training → Contrastive objective: uses objective
- Researcher → LSTM experiment: starts experiment
- Researcher → MLP experiment: starts experiment
- Researcher → Local-window training: starts variant
- Researcher → Index audit: audits indices