Do visual tasks have relationships, or are they unrelated? For instance, could having surface normals simplify estimating the depth of an image? Intuition answers these questions positively, implying existence of a certain structure among visual tasks. Knowing this structure has notable values; it provides a principled way for identifying relationships across tasks, for instance, in order to reuse supervision among tasks with redundancies or solve many tasks in one system without piling up the complexity. We propose a fully computational approach for modeling the transfer learning structure of the space of visual tasks. This is done via finding transfer learning dependencies across tasks in a dictionary of twenty-six 2D, 2.5D, 3D, and semantic tasks. The product is a computational taxonomic map among tasks for transfer learning, and we exploit it to reduce the demand for labeled data. For example, we show that the total number of labeled datapoints needed for solving a set of 10 tasks can be reduced by roughly 2/3 (compared to training independently) while keeping the performance nearly the same. We provide a set of tools for computing and visualizing this taxonomical structure at http://taskonomy.vision.
Figure 1: Two agents in Gibson Environment for real-world perception. The agent is active, embodied, and subject to constraints of physics and space (a,b). It receives a constant stream of visual observations as if it had an on-board camera (c). It can also receive additional modalities, e.g. depth, semantic labels, or normals (d,e,f). The visual observations are from real-world rather than an artificially designed space. AbstractDeveloping visual perception models for active agents and sensorimotor control are cumbersome to be done in the physical world, as existing algorithms are too slow to efficiently learn in real-time and robots are fragile and costly. This has given rise to learning-in-simulation which consequently casts a question on whether the results transfer to real-world. In this paper, we are concerned with the problem of developing real-world perception for active agents, propose Gibson Virtual Environment 1 for this purpose, and showcase sample perceptual tasks learned therein. Gibson is based on virtualizing real spaces, rather than using artificially designed ones, and currently includes over 1400 floor spaces from 572 full buildings. The main characteristics of Gibson are: I. being from the real-world and reflecting its semantic complexity, II. having an internal synthesis mechanism, "Goggles", enabling deploying the trained models in real-world without needing domain adaptation, III. embodiment of agents and making them subject to constraints of physics and space.
Do visual tasks have a relationship, or are they unrelated? For instance, could having surface normals simplify estimating the depth of an image? Intuition answers these questions positively, implying existence of a structure among visual tasks. Knowing this structure has notable values; it is the concept underlying transfer learning and provides a principled way for identifying redundancies across tasks, e.g., to seamlessly reuse supervision among related tasks or solve many tasks in one system without piling up the complexity.We proposes a fully computational approach for modeling the structure of space of visual tasks. This is done via finding (first and higher-order) transfer learning dependencies across a dictionary of twenty six 2D, 2.5D, 3D, and semantic tasks in a latent space. The product is a computational taxonomic map for task transfer learning. We study the consequences of this structure, e.g. nontrivial emerged relationships, and exploit them to reduce the demand for labeled data. For example, we show that the total number of labeled datapoints needed for solving a set of 10 tasks can be reduced by roughly 2 3 (compared to training independently) while keeping the performance nearly the same. We provide a set of tools for computing and probing this taxonomical structure including a solver that users can employ to devise efficient supervision policies for their use cases.
The stability and geometry of a hydrogen-bonded dimer is traditionally attributed mainly to the central moiety A-H⋅⋅⋅B, and is often discussed only in terms of electrostatic interactions. The influence of substituents and of interactions other than electrostatic ones on the stability and geometry of hydrogen-bonded complexes has seldom been addressed. An analysis of the interaction energy in the water dimer and several alcohol dimers--performed in the present work by using symmetry-adapted perturbation theory--shows that the size and shape of substituents strongly influence the stabilization of hydrogen-bonded complexes. The larger and bulkier the substituents are, the more important the attractive dispersion interaction is, which eventually becomes of the same magnitude as the total stabilization energy. Electrostatics alone are a poor predictor of the hydrogen-bond stability trends in the sequence of dimers investigated, and in fact, dispersion interactions predict these trends better.
The asymmetric molybdenum(VI) dioxo complexes of the bis(phenolate) ligands 1,4-bis(2-hydroxybenzyl)-1,4-diazepane, 1,4-bis(2-hydroxy-4-methylbenzyl)-1,4-diazepane, 1,4-bis(2-hydroxy-3,5-dimethylbenzyl)-1,4-diazepane, 1,4-bis(2-hydroxy-3,5-di-tert-butylbenzyl)-1,4-diazepane, 1,4-bis(2-hydroxy-4-flurobenzyl)-1,4-diazepane, and 1,4-bis(2-hydroxy-4-chlorobenzyl)-1,4-diazepane (H(2)(L1)-H(2)(L6), respectively) have been isolated and studied as functional models for molybdenum oxotransferase enzymes. These complexes have been characterized as asymmetric complexes of type [MoO(2)(L)] 1-6 by using NMR spectroscopy, mass spectrometry, elemental analysis, and electrochemical methods. The molecular structures of [MoO(2)(L)] 1-4 have been successfully determined by single-crystal X-ray diffraction analyses, which show them to exhibit a distorted octahedral coordination geometry around molybdenum(VI) in an asymmetrical cis-β configuration. The Mo-O(oxo) bond lengths differ only by ≈0.01 Å. Complexes 1, 2, 5, and 6 exhibit two successive Mo(VI)/Mo(V) (E(1/2), -1.141 to -1.848 V) and Mo(V)/Mo(IV) (E(1/2), -1.531 to -2.114 V) redox processes. However, only the Mo(VI)/Mo(V) redox couple was observed for 3 and 4, suggesting that the subsequent reduction of the molybdenum(V) species is difficult. Complexes 1, 2, 5, and 6 elicit efficient catalytic oxygen-atom transfer (OAT) from dimethylsulfoxide (DMSO) to PMe(3) at 65 °C at a significantly faster rate than the symmetric molybdenum(VI) complexes of the analogous linear bis(phenolate) ligands known so far to exhibit OAT reactions at a higher temperature (130 °C). However, complexes 3 and 4 fail to perform the OAT reaction from DMSO to PMe(3) at 65 °C. DFT/B3LYP calculations on the OAT mechanism reveal a strong trans effect.
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.