Take It or Leave It: Intent-Controlled Partial Optimal Transport
Salil Parth Tripathi, Bertrand Chapron, Fabrice Collard, Nicolas Courty, Ronan Fablet · 2026
Read Take It or Leave It: Intent-Controlled Partial Optimal Transport by Salil Parth Tripathi, Bertrand Chapron, Fabrice Collard, Nicolas Courty, Ronan Fablet (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-20030.txt.
Introduction
Salil Parth Tripathi and four co-workers let partial optimal transport reject by local information. Partial transport leaves mass unmatched by a global budget, rebate or uniform rule, but tasks need pointwise choices tied to reliability or side information. Their intent-controlled version sets pointwise rejection costs on both measures, reads as acceptance thresholds, and solves as a balanced Kantorovich problem on an augmented support. In positive-unlabelled learning and open-partial domain adaptation it improves baselines, and a satellite-ocean case shows sensor priors shaping which signal compares.
Catalog record
| CREATOR | Salil Parth Tripathi, Bertrand Chapron, Fabrice Collard, Nicolas Courty, Ronan Fablet |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 89337 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.20030
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Take It or Leave It: Intent-Controlled Partial Optimal Transport Salil Parth Tripathi, Bertrand Chapron, Fabrice Collard, Nicolas Courty, Ronan Fablet Abstract While optimal transport (OT) enforces a rigid constraint by requiring two measures to be matched exactly, partial optimal transport relaxes this requirement by allowing mass to remain unmatched through a global budget, scalar rebate, or uniform rejection rule. However, many applications call for more structured, pointwise rejection mechanisms, where the decision to leave mass unmatched depends on side-specific reliability, support geometry, or external information about which components should participate in the comparison. We introduce intent-controlled partial optimal transport (IC-POT), a targeted generalization of partial transport that replaces the global rejection paradigm with pointwise rejection costs over both measures. We show that the resulting optimization problem admits a dual interpretation in terms of local acceptance thresholds and can be solved by recasting it as a balanced Kantorovich OT problem on an augmented support. Beyond theoretical analysis, we demonstrate the practical relevance of IC-POT in settings where rejection is driven by side information. In positive-unlabeled learning and open-partial domain adaptation, incorporating pointwise rejection rules that encode statistical structure improves fixed baseline pipelines. Finally, we motivate the use of IC-POT with a geophysical practical case: multi-modal satellite ocean measurements, for which physical and sensors priors naturally inform the rejection mechanism and define the retrieved comparable signal information. 1 Introduction Optimal transport (OT) provides a principled way to compare and align distributions under a transport cost, and has become a standard tool in machine learning, computer vision, and scientific computing ( 28 ) . In its classical balanced form, all source mass must be transported and all target mass must be explained. This full-participation assumption is often too rigid and may result in artificial correspondences, negative transfer, or spurious matches between structures that should play different roles in the comparison. Partial and unbalanced OT introduce non-participation by allowing some mass to remain unmatched ( 4 ; 6 ; 23 ; 8 ) . In most formulations, however, this choice is governed by a global transported-mass budget, a scalar rebate, or a marginal penalty. These mechanisms only control how much mass is rejected, not which points should be protected, exposed, or made inexpensive to discard. Many applications require a formulation in which leaving mass unmatched can be priced locally, according to the information available on each support. In positive-unlabeled learning, rejection is tied to the sampling mechanism ( 6 ) . In open-partial domain adaptation, it is tied to confidence and target geometry ( 9 ; 5 ; 25 ) . In geophysical comparison, the question is often not whether two observations differ, but which parts of them can be meaningfully compared given their measurement geometry and sensor-specific uncertainties ( 17 ; 16 ; 33 ) . To address these challenges, we introduce intent-controlled partial optimal transport (IC-POT), a targeted extension of partial OT formulated with explicit source-side and target-side slack variables. Pointwise costs attached to these slacks price the local alternative of leaving mass unmatched, while the pairwise transport cost continues to describe the geometry of matching. The modeling shift is simple: the unmatched policy becomes representable directly on the original supports. We show that this formulation yields an equivalent Lagrangian sub-coupling problem in the sense of Caffarelli–McCann ( 4 ) , with the scalar transport rebate replaced by a separable local rebate, one term on each marginal. This reduced form preserves the side-specific interpretation of the unmatched costs and leads to dual acceptance caps, an exact admissibility rule for active transport edges, and a strict extension of the constant-cost partial OT model. Experiments on positive-unlabeled learning, open-partial domain adaptation, and multimodal satellite-derived ocean-wave spectra show how IC-POT leverages side information to specify the unmatched policy. PU learning and OPDA test statistical settings where sampling bias, confidence, and latent geometry determine which samples should remain protected from rejection. The ocean-wave experiment compares spectra retrieved from SWIM, the wave scatterometer onboard CFOSAT ( 17 ) , and Synthetic Aperture Radar (SAR) ( 16 ) , and illustrates a distinct role of the formulation: physical priors naturally inform the rejection mechanism and define the comparable signal itself. Across these experiments, the transport geometry and baseline pipeline are kept fixed; what changes is the way unmatched mass is priced. Overall, the contribution of the paper is threefold. First, it makes unmatched behavior representable as an explicit object in the formulation through a slack formulation on the original support. Second, it characterizes the resulting problem through an equivalent separable pointwise-rebate generalization of the classical Lagrangian partial transport, with Kantorovich equivalence, dual caps, sparse admissibility, and strict separation from constant-cost partial OT. Third, it shows empirically that this added modeling capacity matters in reject-structured tasks compared with constant-cost partial OT. 2 Background and Related Work Optimal transport compares measures under a pairwise transport cost ( 28 ; 34 ) . In the discrete balanced Kantorovich setting, we consider finite supports X=\{x_{i}\}_{i=1}^{n} , Y=\{y_{j}\}_{j=1}^{m} , masses \mu\in\mathbb{R}_{+}^{n} , \nu\in\mathbb{R}_{+}^{m} , and a transport cost matrix C\in\mathbb{R}_{+}^{n\times m} . The classical problem searches for a coupling P\in\mathbb{R}_{+}^{n\times m} solving \min_{P\geq 0}\ \langle C,P\rangle\qquad\text{s.t.}\qquad P\mathbf{1}_{m}=\mu,\quad P^{\top}\mathbf{1}_{n}=\nu. (1) This full-matching requirement is often too rigid when part of one support should remain unmatched. Classical partial OT addresses this issue by optimizing over sub-couplings. In the Lagrangian formulation of Caffarelli–McCann, the amount of transported mass is governed by a scalar rebate: transporting one more unit is rewarded uniformly, independently of where it comes from or where it goes ( 4 ) . The constant-cost construction of 6 is a discrete version of the same global-control regime. Unbalanced OT relaxes the marginal constraints through global penalties ( 23 ; 8 ) , while semi-relaxed OT allows one-sided marginal relaxation natively ( 15 ) . Recent contributions such as mini-batch partial transportation ( 26 ) and progressive partial OT ( 35 ) mainly improve optimization or scheduling while keeping the same type of global reject control. There is also a broader line of work on constrained and sparse OT frameworks. Capacity and other feasibility constraints modify the admissible set of couplings directly ( 19 ; 11 ) , while smooth, sparse, and sparsity-constrained OT aim to structure the transport plan itself through regularization or explicit support restrictions ( 2 ; 24 ) . Through related to our work, these formulations operate on a different object: they shape feasible couplings or the geometry of the domain, whereas we aim to make unmatched behavior itself dependent on pointwise structure and side information while keeping a partial OT framework. Several application areas already require control over which samples participate in transport. Guided color or style transfer uses masks, semantic constraints, or user guidance to preserve selected regions ( 14 ; 22 ; 18 ) . In domain adaptation, OT has been used to align domains and infer transferable structure from the transport plan itself ( 9 ; 5 ) , while masked OT has been proposed in partial domain adaptation because standard OT assumptions can induce biased alignments and negative transfer ( 25 ) . In selective matching and support subset selection, fully relaxed or globally penalized formulations may suppress points that should remain active in the matching