Multi-task Learning with Labeled and Unlabeled Tasks Supplementary Material
نویسندگان
چکیده
1. Preliminaries In this section we list a few results from the literature that will be utilized in the proof of Theorem 1. Proposition 1 (Lemma 1 in (Ben-David et al., 2010)). Let d be the VC dimension of the hypothesis set H and S1, S2 be two i.i.d. samples of size n from D1 and D2 respectively. Then for any δ > 0 with probability at least 1− δ: disc(D1, D2)≤disc(S1, S2)+2 √ 2d log(2n) + log(2/δ) n .
منابع مشابه
Multi-task Learning with Labeled and Unlabeled Tasks
In multi-task learning, a learner is given a collection of prediction tasks and needs to solve all of them. In contrast to previous work, which required that annotated training data is available for all tasks, we consider a new setting, in which for some tasks, potentially most of them, only unlabeled training data is provided. Consequently, to solve all tasks, information must be transferred b...
متن کاملSupplementary material: Learning and Selecting Features viaPoint-wise Gated Boltzmann Machines
There are many classification tasks where we are given a large number of unlabeled examples in addition to only a few labeled training examples. For such scenario, it is important to include unlabeled examples during the training to generalize well to the unseen data, and thus avoid overfitting. Larochelle and Bengio (2008) proposed the semi-supervised training of the discriminative restricted ...
متن کاملConstraint-Driven Active Learning Across Tasks
Consider a set of T tasks, each with a response variable Yi, i = 1, 2, . . . , T . Each sample in our training set x ∈ U is associated with these T labels (labels are not necessarily binary). For each sample, we might know a few (or none) of its T labels. We use UL(x) to denote the set of unknown labels on a sample x: UL(x) = {Yi : Yi is unknown for sample x}. Given labeled examples, we can lea...
متن کاملDeceptive Review Spam Detection via Exploiting Task Relatedness and Unlabeled Data
Existing work on detecting deceptive reviews primarily focuses on feature engineering and applies off-the-shelf supervised classification algorithms to the problem. Then, one real challenge would be to manually recognize plentiful ground truth spam review data for model building, which is rather difficult and often requires domain expertise in practice. In this paper, we propose to exploit the ...
متن کاملActive Task Selection for Multi-Task Learning
In multi-task learning, a learner is given a collection of prediction tasks and needs to solve all of them. In contrast to previous work, which required that annotated training data is available for all tasks, we consider a new setting, in which for some tasks, potentially most of them, only unlabeled training data is provided. Consequently, to solve all tasks, information must be transferred b...
متن کامل