Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384111941> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4384111941 abstract "Deep Boltzmann machines (DBMs), one of the first ``deep'' learning methods ever studied, are multi-layered probabilistic models governed by a pairwise energy function that describes the likelihood of all variables/nodes in the network. In practice, DBMs are often constrained, i.e., via the emph{restricted} Boltzmann machine (RBM) architecture (which does not permit intra-layer connections), in order to allow for more efficient inference. In this work, we revisit the generic DBM approach, and ask the question: are there other possible restrictions to their design that would enable efficient (approximate) inference? In particular, we develop a new class of restricted model, the monotone DBM, which allows for arbitrary self-connection in each layer, but restricts the emph{weights} in a manner that guarantees the existence and global uniqueness of a mean-field fixed point. To do this, we leverage tools from the recently-proposed monotone Deep Equilibrium model and show that a particular choice of activation results in a fixed-point iteration that gives a variational mean-field solution. While this approach is still largely conceptual, it is the first architecture that allows for efficient approximate inference in fully-general weight structures for DBMs. We apply this approach to simple deep convolutional Boltzmann architectures and demonstrate that it allows for tasks such as the joint completion and classification of images, within a single deep probabilistic setting, while avoiding the pitfalls of mean-field inference in traditional RBMs." @default.
- W4384111941 created "2023-07-13" @default.
- W4384111941 creator A5037951424 @default.
- W4384111941 creator A5075035644 @default.
- W4384111941 creator A5075312638 @default.
- W4384111941 date "2023-07-10" @default.
- W4384111941 modified "2023-09-24" @default.
- W4384111941 title "Monotone deep Boltzmann machines" @default.
- W4384111941 doi "https://doi.org/10.48550/arxiv.2307.04990" @default.
- W4384111941 hasPublicationYear "2023" @default.
- W4384111941 type Work @default.
- W4384111941 citedByCount "0" @default.
- W4384111941 crossrefType "posted-content" @default.
- W4384111941 hasAuthorship W4384111941A5037951424 @default.
- W4384111941 hasAuthorship W4384111941A5075035644 @default.
- W4384111941 hasAuthorship W4384111941A5075312638 @default.
- W4384111941 hasBestOaLocation W43841119411 @default.
- W4384111941 hasConcept C108583219 @default.
- W4384111941 hasConcept C11413529 @default.
- W4384111941 hasConcept C126255220 @default.
- W4384111941 hasConcept C134306372 @default.
- W4384111941 hasConcept C153083717 @default.
- W4384111941 hasConcept C154945302 @default.
- W4384111941 hasConcept C155846161 @default.
- W4384111941 hasConcept C184898388 @default.
- W4384111941 hasConcept C192576344 @default.
- W4384111941 hasConcept C199354608 @default.
- W4384111941 hasConcept C2524010 @default.
- W4384111941 hasConcept C2776214188 @default.
- W4384111941 hasConcept C2777021972 @default.
- W4384111941 hasConcept C2834757 @default.
- W4384111941 hasConcept C33923547 @default.
- W4384111941 hasConcept C41008148 @default.
- W4384111941 hasConcept C49937458 @default.
- W4384111941 hasConcept C80444323 @default.
- W4384111941 hasConceptScore W4384111941C108583219 @default.
- W4384111941 hasConceptScore W4384111941C11413529 @default.
- W4384111941 hasConceptScore W4384111941C126255220 @default.
- W4384111941 hasConceptScore W4384111941C134306372 @default.
- W4384111941 hasConceptScore W4384111941C153083717 @default.
- W4384111941 hasConceptScore W4384111941C154945302 @default.
- W4384111941 hasConceptScore W4384111941C155846161 @default.
- W4384111941 hasConceptScore W4384111941C184898388 @default.
- W4384111941 hasConceptScore W4384111941C192576344 @default.
- W4384111941 hasConceptScore W4384111941C199354608 @default.
- W4384111941 hasConceptScore W4384111941C2524010 @default.
- W4384111941 hasConceptScore W4384111941C2776214188 @default.
- W4384111941 hasConceptScore W4384111941C2777021972 @default.
- W4384111941 hasConceptScore W4384111941C2834757 @default.
- W4384111941 hasConceptScore W4384111941C33923547 @default.
- W4384111941 hasConceptScore W4384111941C41008148 @default.
- W4384111941 hasConceptScore W4384111941C49937458 @default.
- W4384111941 hasConceptScore W4384111941C80444323 @default.
- W4384111941 hasLocation W43841119411 @default.
- W4384111941 hasOpenAccess W4384111941 @default.
- W4384111941 hasPrimaryLocation W43841119411 @default.
- W4384111941 hasRelatedWork W2047776774 @default.
- W4384111941 hasRelatedWork W2240647871 @default.
- W4384111941 hasRelatedWork W2567271240 @default.
- W4384111941 hasRelatedWork W2892911634 @default.
- W4384111941 hasRelatedWork W2955124940 @default.
- W4384111941 hasRelatedWork W2963058055 @default.
- W4384111941 hasRelatedWork W2971852391 @default.
- W4384111941 hasRelatedWork W3044458868 @default.
- W4384111941 hasRelatedWork W3082895349 @default.
- W4384111941 hasRelatedWork W870929296 @default.
- W4384111941 isParatext "false" @default.
- W4384111941 isRetracted "false" @default.
- W4384111941 workType "article" @default.