Online Dictionary Learning with Group Structure Inducing Norms
نویسندگان
چکیده
• Sparse coding. • Structured sparsity (e.g., disjunct groups, trees): increased performance in several applications. • Our goal: develop a dictionary learning method, which – enables general overlapping group structures, – is online: fast, memory efficient, adaptive, – applies non-convex sparsity inducing regularization: ∗ fewer measurements, ∗ weaker conditions on the dictionary, ∗ robust (w.r.t. noise, compressibility). – can deal with missing information. Current approaches can exhibit two of these features at most.
منابع مشابه
Task-Induced Involvement in L2 Vocabulary Learning: A Case for Listening Comprehension
The study aimed at investigating whether the retention of vocabulary acquired incidentally is dependent upon the amount of task-induced involvement. Immediate and delayed retention of twenty unfamiliar words was examined in three learning tasks( listening comprehension + group discussion, listening comprehension + dictionary checking + summary writing in L1, and listening comprehension + dictio...
متن کاملSpeech Enhancement using Adaptive Data-Based Dictionary Learning
In this paper, a speech enhancement method based on sparse representation of data frames has been presented. Speech enhancement is one of the most applicable areas in different signal processing fields. The objective of a speech enhancement system is improvement of either intelligibility or quality of the speech signals. This process is carried out using the speech signal processing techniques ...
متن کاملSparse Hierachical Extrapolated Parametric Methods for Cortical Data Analysis
Many neuroimaging studies focus on the cortex, in order to benefit from better signal to noise ratios and reduced computational burden. Cortical data are usually projected onto a reference mesh, where subsequent analyses are carried out. Several multiscale approaches have been proposed for analyzing these surface data, such as spherical harmonics and graph wavelets. As far as we know, however, ...
متن کاملLearning Hierarchical and Topographic Dictionaries with Structured Sparsity
Recent work in signal processing and statistics have focused on defining new regularization functions, which not only induce sparsity of the solution, but also take into account the structure of the problem. We present in this paper a class of convex penalties introduced in the machine learning community, which take the form of a sum of l2and l∞-norms over groups of variables. They extend the c...
متن کاملConcordance-Based Data-Driven Learning Activities and Learning English Phrasal Verbs in EFL Classrooms
In spite of the highly beneficial applications of corpus linguistics in language pedagogy, it has not found its way into mainstream EFL. The major reasons seem to be the teachers’ lack of training and the unavailability of resources, especially computers in language classes. Phrasal verbs have been shown to be a problematic area of learning English as a foreign language due to their semantic op...
متن کامل