Multitask Sparsity via Maximum Entropy Discrimination

By Tony Jebara. 2011

72 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

A Survey on Multi-Task Learning (2017, neutral): “Similar to [34], Jebara [37], [38] proposes to learn a binary indicator vector to do multi-task feature selection based on”
Multi-task and Lifelong Learning of Kernels (2015, methodology): “A natural extension of representation learning approach was proposed for kernel methods in [12,13], where the authors described a method for learning a kernel that is shared between tasks as a combination of some base kernels using maximum entropy discrimination approach.”
Infinite Latent SVM for Classification and Multi-task Learning (2011, methodology): “Moreover, different from [16], which learns a binary vector with known dimensionality to select features or kernels on x, we learn an unbounded projection matrix Z using nonparametric Bayesian techniques.”
Sparse Bayesian Multi-Task Learning (2011, background): “This approach views the multi-task learning problem as a transfer learning problem, where it is assumed that the various tasks belong in some sense to the same environment and share common properties [23, 5].”
Feature Selection Based on Structured Sparsity: A Comprehensive Study (2017, methodology): “A multitask feature selection method via maximum entropy discrimination (MED) was proposed in [114], and the experimental results show that MED multitask learning outperforms single-task”
On Better Exploring and Exploiting Task Relationships in Multitask Learning: Joint Model and Feature Learning (2018, methodology): “As one of the most popular single-task learning methods, a support vector machine (SVM) has been studied in many MTL works [5], [12], [25]–[30].”
Approximate inference for the loss-calibrated Bayesian (2011, unknown): “Following the more modern treatment of Jebara (2011), MED aims at solving the following optimization problem (using our notation): qMED = arg min q∈Q KL(q(θ)||p(θ)) + C ∑ i ξi (29) s.t. ξi + pq(yi|xi) ≥ pq(y|xi) + `(yi, y) ∀i, y ∈ Y, though in practice they use ∫ Θ q(θ) log p(y|x, θ)dθ rather than…”