Principled approaches for learning to defer with multiple experts

A Mao, M Mohri, Y Zhong - International Workshop on Combinatorial …, 2024 - Springer
We present a study of surrogate losses and algorithms for the general problem of learning to
defer with multiple experts. We first introduce a new family of surrogate losses specifically …

Multi-label learning with stronger consistency guarantees

A Mao, M Mohri, Y Zhong - arxiv preprint arxiv:2407.13746, 2024 - arxiv.org
We present a detailed study of surrogate losses and algorithms for multi-label learning,
supported by $ H $-consistency bounds. We first show that, for the simplest form of multi …

A universal growth rate for learning with smooth surrogate losses

A Mao, M Mohri, Y Zhong - arxiv preprint arxiv:2405.05968, 2024 - arxiv.org
This paper presents a comprehensive analysis of the growth rate of $ H $-consistency
bounds (and excess error bounds) for various surrogate losses used in classification. We …

Cardinality-Aware Set Prediction and Top- Classification

C Cortes, A Mao, C Mohri, M Mohri, Y Zhong - arxiv preprint arxiv …, 2024 - arxiv.org
We present a detailed study of cardinality-aware top-$ k $ classification, a novel approach
that aims to learn an accurate top-$ k $ set predictor while maintaining a low cardinality. We …

-Consistency Guarantees for Regression

A Mao, M Mohri, Y Zhong - arxiv preprint arxiv:2403.19480, 2024 - arxiv.org
We present a detailed study of $ H $-consistency bounds for regression. We first present
new theorems that generalize the tools previously given to establish $ H $-consistency …

Enhanced -Consistency Bounds

A Mao, M Mohri, Y Zhong - arxiv preprint arxiv:2407.13722, 2024 - arxiv.org
Recent research has introduced a key notion of $ H $-consistency bounds for surrogate
losses. These bounds offer finite-sample guarantees, quantifying the relationship between …

Realizable -Consistent and Bayes-Consistent Loss Functions for Learning to Defer

A Mao, M Mohri, Y Zhong - arxiv preprint arxiv:2407.13732, 2024 - arxiv.org
We present a comprehensive study of surrogate loss functions for learning to defer. We
introduce a broad family of surrogate losses, parameterized by a non-increasing function …

Learning to Partially Defer for Sequences

S Rayan, A Tewari - arxiv preprint arxiv:2502.01459, 2025 - arxiv.org
In the Learning to Defer (L2D) framework, a prediction model can either make a prediction or
defer it to an expert, as determined by a rejector. Current L2D methods train the rejector to …

Learning to Help in Multi-Class Settings

Y Wu, Y Li, Z Dong, N Sathyavageeswaran… - arxiv preprint arxiv …, 2025 - arxiv.org
Deploying complex machine learning models on resource-constrained devices is
challenging due to limited computational power, memory, and model retrainability. To …

Domain Adaptation for Robust Model Routing

C Dann, Y Mansour, TV Marinov, M Mohri - Adaptive Foundation Models … - openreview.net
The rapid proliferation of domain-specialized machine learning models presents a
challenge: while individual models excel in specific domains, their performance varies …