Add transductive mode support and narrow ml_selector bounds#75
Merged
Add transductive mode support and narrow ml_selector bounds#75
Conversation
Collaborator
johanos1
commented
Jan 27, 2026
johanos1
commented
- TorchClient/TorchGeometricClient: Handle transductive mode with train/val/test masks
- SklearnClient: Fix label column extraction to use 'is_sar' explicitly
- utils.py: Fix tensordatasets/graphdataset to extract labels by column name
- config.py: Fix build_data_paths for TorchClient (trainset vs trainset_nodes)
- models.yaml: Tune GNN hyperparameters (dropout 0.9→0.3, lr, hidden_dim)
- data.yaml: Narrow structure_weights bounds to [-0.1, 0.1] to reduce centrality bias
- Training modules: Support transductive mode path handling
- Tests: Update for new transductive mode behavior
- TorchClient/TorchGeometricClient: Handle transductive mode with train/val/test masks - SklearnClient: Fix label column extraction to use 'is_sar' explicitly - utils.py: Fix tensordatasets/graphdataset to extract labels by column name - config.py: Fix build_data_paths for TorchClient (trainset vs trainset_nodes) - models.yaml: Tune GNN hyperparameters (dropout 0.9→0.3, lr, hidden_dim) - data.yaml: Narrow structure_weights bounds to [-0.1, 0.1] to reduce centrality bias - Training modules: Support transductive mode path handling - Tests: Update for new transductive mode behavior
Problem: SAR accounts had perfect separators (min_gap=0, burstiness=-inf) regardless of burstiness_bias_alert parameter settings. Root causes: 1. scatter_gather, gather_scatter, and stack patterns used randint() directly, bypassing PatternScheduler and ignoring burstiness params 2. PatternScheduler.get_transaction_steps() used .astype(int) which caused multiple transactions to land on same integer step Solution: - Refactored scatter_gather, gather_scatter, stack to use PatternScheduler.get_transaction_steps() with phase ordering - Added unique step distribution in get_transaction_steps(): when num_transactions <= period, spread to unique integer steps using beta-weighted sampling without replacement Also fixes preprocessor plotting: - Use linear scale centered at 1.0x instead of log2 - Handle near-zero denominators and sign differences in ratio calc
Tests verify that client_type is read from config file (default.client_type) rather than passed as a parameter, matching the current implementation.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.