Corpus ID: 7943096

Regret Bounds for Non-decomposable Metrics with Missing Labels

@inproceedings{Natarajan2016RegretBF,
  title={Regret Bounds for Non-decomposable Metrics with Missing Labels},
  author={N. Natarajan and P. Jain},
  booktitle={NIPS},
  year={2016}
}
  • N. Natarajan, P. Jain
  • Published in NIPS 2016
  • Computer Science, Mathematics
  • We consider the problem of recommending relevant labels (items) for a given data point (user). In particular, we are interested in the practically important setting where the evaluation is with respect to non-decomposable (over labels) performance metrics like the $F_1$ measure, and the training data has missing labels. To this end, we propose a generic framework that given a performance metric $\Psi$, can devise a regularized objective function and a threshold such that all the values in the… CONTINUE READING
    On the benefits of output sparsity for multi-label classification
    5
    Rank Pruning Approach for Noisy Multi-label Learning

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 23 REFERENCES
    Large-scale Multi-label Learning with Missing Labels
    314
    PU Learning for Matrix Completion
    78
    Surrogate regret bounds for bipartite ranking via strongly proper losses
    38
    Consistent Multilabel Ranking through Univariate Losses
    25
    Surrogate regret bounds for generalized classification performance metrics
    12
    On the Consistency of Multi-Label Learning
    99
    Sparse Local Embeddings for Extreme Multi-label Classification
    248
    Provable Inductive Matrix Completion
    109
    Low Rank Matrix Completion with Exponential Family Noise
    36
    Consistent Multilabel Classification
    41