Here we present DeepGaze II, a model that predicts where people look in images. The model uses the features from the VGG-19 deep neural network trained to identify objects in images. Contrary to other saliency models that use deep features, here we use the VGG features for saliency prediction with no additional fine-tuning (rather, a few readout layers are trained on top of the VGG features to predict saliency). The model is therefore a strong test of transfer learning. After conservative crossvalidation, DeepGaze II explains about 87% of the explainable information gain in the patterns of fixations and achieves top performance in area under the curve metrics on the MIT300 hold-out benchmark. These results corroborate the finding from DeepGaze I (which explained 56% of the explainable information gain), that deep features trained on object recognition provide a versatile feature space for performing related visual tasks. We explore the factors that contribute to this success and present several informative image examples. A web service is available to compute model predictions at http://deepgaze.bethgelab.org.
Summary Rauwolfia of Indian commerce consists of the dried rhizomes and roots, with occasional small pieces of attached aerial stem, of Rauwolfia serpentina Benth. The important histological features are the cork, composed of alternating layers of broad and narrow cells, giving a somewhat spongy and friable exterior to the drug; the narrow bark and the wide central mass of wood in the root or broad ring in the rhizome, showing about 3 to 8 growth rings; a tetrarch, or occasionally triarch, primary xylem in roots or a very small pith at the centre of rhizomes; in the periphery of the pith a ring of about 20 small groups of perimedullary phloem. All the cells of the wood, including those of the medullary rays are lignified. The xylem vessels, fibres and parenchyma form narrow undulating radially arranged bands, separated by large medullary rays in the root and by smaller ones in the rhizome. The vessel segments are small and narrow and the xylem fibres are often irregularly shaped with long, usually much contorted, tapering ends; many of the vessels, especially in the rhizome, show tylosis and occasionally contain resinous masses. The phloem also occurs in undulating bands similar to those of the xylem; phloem fibres are absent. In the rhizome, unlignified pericyclic fibres occur, either singly or in groups of 2 to 4; they show the local and sub-terminal elongated-oval enlargements characteristic of the Apocynaceæ. All the parenchymatous tissues, except the cork, contain numerous starch grains, about 4 to 20 to 50μ in diameter. Cells containing a brownish yellow secretion occur in small numbers, either singly or in short longitudinal rows, in the bark, wood and pith, being most frequent in the cortex of pieces of rhizome. Typical laticiferous tubes are absent. Calcium oxalate occurs in numerous cells of the phloem and of the medullary rays of the bark; the crystals are either well-formed prisms or aggregates of irregular angular components.
In this paper we contend that the engineering of information systems is hampered by a paucity of tools to tractably model, simulate and predict the impact of realistic user behaviours on the emergent properties of the wider socio-technical system, evidenced by the plethora of case studies of system failure in the literature. We address this gap by presenting a novel approach that models ideal user behaviour as workflows, and introduces irregularities in that behaviour as aspects which fuzz the model. We demonstrate the success of this approach through a case study of software development workflows, showing that the introduction of realistic user behaviour to idealised workflows better simulates outcomes reported in the empirical software engineering literature.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.