SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2509.24100 · ICML · 2025

SpeedCP: Fast Kernel-Based Conditional Conformal Prediction

Zixuan Wu, Yating Liu, Jin Yeo, Jung, So Jeong, Claire Donnat

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 15 functions out of this paper's own repositories and ran 0 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
yeojin-jung/speedcp canonical 0 of 15
FunctionStatusWhere it lives
add_node_feats Not yet run yeojin-jung/speedcp/experiments/common/moleculenet_helpers.py
code served (permissive licence) · get_code("07af09a57ea1d5b5")
clr Not yet run yeojin-jung/speedcp/speedcp/utils.py
code served (permissive licence) · get_code("30de34a50a7281c6")
compute_coverage Not yet run yeojin-jung/speedcp/experiments/common/moleculenet_helpers.py
code served (permissive licence) · get_code("562210c7f6aec45c")
fcp_predict_inner Not yet run yeojin-jung/speedcp/experiments/real_data/run_mri.py
code served (permissive licence) · get_code("1f48dc69f99d66e4")
fcp_scores Not yet run yeojin-jung/speedcp/experiments/real_data/run_mri.py
code served (permissive licence) · get_code("fd151645cecab3f8")
kernel Not yet run yeojin-jung/speedcp/speedcp/utils.py
code served (permissive licence) · get_code("62ef2c7c0e6f7e9d")
lambda_init Not yet run yeojin-jung/speedcp/speedcp/lambda_trace.py
code served (permissive licence) · get_code("c643ffaaf3eaece0")
load_or_fit_reference_pca Not yet run yeojin-jung/speedcp/experiments/molecular/run_esol.py
code served (permissive licence) · get_code("a910623f46443974")
load_or_fit_reference_pca Not yet run yeojin-jung/speedcp/experiments/molecular/run_qm7b.py
code served (permissive licence) · get_code("cf54cfcea83766df")
load_or_fit_reference_pca Not yet run yeojin-jung/speedcp/experiments/molecular/run_qm9.py
code served (permissive licence) · get_code("a49dada588c51ca8")
make_loader Not yet run yeojin-jung/speedcp/experiments/molecular/run_esol.py
code served (permissive licence) · get_code("a0a653b31952f3d9")
onehot Not yet run yeojin-jung/speedcp/experiments/molecular/run_qm7b.py
code served (permissive licence) · get_code("29f2999901c7b6d6")
pinball Not yet run yeojin-jung/speedcp/speedcp/utils.py
code served (permissive licence) · get_code("49d2ca0ba2052164")
select_target Not yet run yeojin-jung/speedcp/experiments/common/moleculenet_helpers.py
code served (permissive licence) · get_code("5692fbd2e6078394")
vanilla_scores Not yet run yeojin-jung/speedcp/experiments/real_data/run_mri.py
code served (permissive licence) · get_code("2e2e6b1fb2dc45b5")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Conformal prediction provides distribution-free prediction sets with finite-sample conditional guarantees. RKHS-based frameworks-while promising for complex covariate shifts-suffer from prohibitive computational costs. To guarantee conditional validity under such shifts while ensuring feasibility, we build upon the framework of Gibbs et al. [2025] by introducing a stable and efficient algorithm that computes the full solution path of the regularized RKHS conformal optimization problem, at essentially the same cost as a single kernel quantile fit. Our approach provides simultaneous hyperparameter tuning for smoothness control and data-adaptive calibration. To extend the method to high-dimensional settings, we further integrate our approach with low-rank latent embeddings that capture conditional validity in a data-driven latent space. Empirically, our method provides reliable conditional coverage across a variety of modern black-box predictors, improving the interval length of Gibbs et al. [2025] by 30%, while achieving a 40-fold speedup.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2509.24100")
get_code_for_paper("2509.24100")
have("2509.24100")

Connect an agent — have() is free.