40,362 research outputs found
Metadata Representations for Queryable ML Model Zoos
No full textMachine learning (ML) practitioners and organizations are building model zoos of pre-trained models, containing metadata describing properties of the ML models and datasets that are useful for reporting, auditing, reproducibility, and interpretability purposes. The metatada is currently not standardised; its expressivity is limited; and there is no interoperable way to store and query it. Consequently, model search, reuse, comparison, and composition are hindered. In this paper, we advocate for standardized ML model metadata representation and management, proposing a toolkit supported to help practitioners manage and query that metadata.Web Information SystemsHuman-Centred Artificial Intelligenc
A Manifesto of Nodalism
Get PDFThis paper proposes the notion of Nodalism as a means describing contemporary culture and of understanding my own creative practice in electronic music composition. It draws on theories and ideas from Kirby, Bauman, Bourriaud, Deleuze, Guatarri, and Gochenour, to demonstrate how networks of ideas or connectionist neural models of cognitive behaviour can be used to contextualize, understand and become a creative tool for the creation of contemporary electronic music
Stimulating designers creativity based on a creative evolutionary system and collective intelligence in product design
No full textIn the field of industry design, the affective response characteristic can plays a significant role that could grasp customer's attention in product. Especially, the design strategy of "affection-prerequisite" shows the importance of the trend of customer-oriented for the industry products. Up to date, in consideration of the rapidly life-cycle on industrial design field, however, how designers can retrieve customer's affective response promptly and using such information into their design work efficiently are significant, but not discussed yet. Based on this concept, the objective of this research is to present a creative stimulation system for designers. The "wisdom of crowds" was collected via the Web in order to ascertain customers' affective responses to product shapes. This data was used to create an evolutionary design system platform by using a design alternatives generation mechanism. The study involved integrating an interactive genetic algorithm into the mechanism to generate an interactive creative stimulation system. An affective design of the shape of a mobile phone was used as an example of implementation and also used as a proof of the research concept. Relevance to industry: HIGA and ESC methods proposed in this research provide the usage of the affective evaluation system for acquiring the customers' affective responses in the conceptual design process of a mobile phone. This design-generation process allows the design stakeholders to realize their customers' affective responses interactively with sharing floating data among them according to the daily update of customer's evaluations. (C) 2009 Elsevier B.V. All rights reserved
Optimizing ML Inference Queries Under Constraints
No full textThe proliferation of pre-trained ML models in public Web-based model zoos facilitates the engineering of ML pipelines to address complex inference queries over datasets and streams of unstructured content. Constructing optimal plan for a query is hard, especially when constraints (e.g. accuracy or execution time) must be taken into consideration, and the complexity of the inference query increases. To address this issue, we propose a method for optimizing ML inference queries that selects the most suitable ML models to use, as well as the order in which those models are executed. We formally define the constraint-based ML inference query optimization problem, formulate it as a Mixed Integer Programming (MIP) problem, and develop an optimizer that maximizes accuracy given constraints. This optimizer is capable of navigating a large search space to identify optimal query plans on various model zoos.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Web Information SystemsHuman-Centred Artificial Intelligenc
FIRST-PRINCIPLES CALCULATIONS OF THE COULOMB PSEUDOPOTENTIAL MU-ASTERISK - APPLICATION TO AL
No full textWe present first-principles pseudopotential calculations for the Coulomb pseudopotential mu* using a dielectric matrix approach. An application to Al is made and mu* is estimated to be 0.091. A comparison of the crystal potential, local field, and exchange-correlation effects on the dielectric screening demonstrates that the exchange-correlation effect contributes the largest reduction in mu* compared to that obtained for a free electron gas
Won Il Lee
No full text학위논문(석사)--아주대학교 일반대학원 :의학과,2015. 2Background: We investigated the correction methods following wrong-settings of emulsion
concentrations of propofol as a countermeasure against erroneous target-controlled infusions
(TCI).
Methods: TCIs were started with targeting 4.0 μg/ml of effect-site concentration (Ceff) of
propofol, and the emulsion concentrations were selected for 2.0% instead of 1.0% (FALSE1-
2, n = 24), or 1.0% instead of 2.0% (FALSE2-1, n = 24). These wrong TCIs were corrected at
3 min after infusion start. During FALSE1-2, the deficit was filled up while injecting after
equilibrium (n = 12), or while overriding (n = 12). During FALSE2-1, the overdose was
evacuated while targeting Ceff (n = 12) or targeting plasma concentration (Cp) (n = 12). The
gravimetrical measurements of TCI reproduced the Cp and Ceff using simulations. The
reproduced Ceff at 3 min (Ceff-3min) and the time to be normalized within ± 5% of target Ceff
(T±5%), were compared between the correction methods.
Results: During the wrong TCI, Ceff-3min was 1.98 ± 0.01 μg/ml in FALSE1-2, and 7.99 ± 0.05
μg/ml in FALSE2-1. In FALSE1-2, T±5% was significantly shorter when corrected while
overriding (3.9 ± 0.25 min), than corrected after equilibrium (6.9 ± 0.05 min) (P < 0.001). In
FALSE2-1, T±5% was significantly shorter during targeting Cp (3.6 ± 0.04 min) than targeting
Ceff (6.7 ± 0.15 min) (P < 0.001).
Conclusions: The correction methods, based on the pharmacokinetic and pharmacodynamic
characteristics, could effectively and rapidly normalize the wrong TCI following erroneously
selections of the emulsion concentration of propofol.Ⅰ.INTRODUCTION 1
Ⅱ.MATERIALS AND METHODS 2
Ⅲ.RESULTS 7
Ⅳ.DISCUSSION 11
Ⅴ.CONCLUSION 15Maste
A polarimetric current sensor using an orthogonally polarized dual-frequency fibre laser
No full textWe demonstrate a novel Er3+-doped fibre laser operating in orthogonally polarized dual-frequency modes and its application to electric current sensing with frequency read-out. A Faraday rotating mirror and spatial hole burning effects in a gain medium and in a saturable absorber are utilized to control the lasing mode and polarization. The polarization mode beat frequency changes linearly in response to the non-reciprocal circular birefringence induced by an external magnetic field. When the laser was applied to sensing an alternating electric current with a simple phase-locked loop signal processing scheme, the slope coefficient of 8.0 kHz A(p-p)(-1), per turn and the noise equivalent current of 460 mu A(rms) HZ(-1/2) per turn were obtained. The new current sensor is immune to perturbations in the lead fibres
Risk assessment for the occurrence of <i>Escherichia coli</i> 0157:H7 in indigenous fermented milk (Lee Naga a Agbora) produced in Uganda
No full textDifferent types of indigenous fermented milks are widely consumed in pastoralist communities of Africa, including Uganda. Local processing of milk generally takes place under uncontrolled hygienic conditions. Lee naga a agbora, a locally fermented milk product in northern Uganda, was evaluated for microbial contamination. The changes in pH, Total Plate Counts (TPC), Total Coliforms (TC), yeasts and moulds, and Escherichia coli 0157:H7, particularly its survival during different stages of fermentation were studied. The pH was found to decrease with fermentation time (6.54 to 4.54). E. coli was detected in most samples of Lee naga a agbora and its occurrence was observed throughout the fermentation period. The mean value of TPC (6.60± 0.10 log cfu ml-1) at the start of fermentation was significantly different (p0.05) between 48 and 72 hours. The rate of decrease in TC counts in the initial fermentation period (0 to 48 hours) was lower than the later stages of fermentation (48 to 144 hours). Yeast and mould counts increased with fermentation time (6.20 to 7.42 log cfu ml-1). On the whole, there is risk associated with consumption of Lee naga a agbora given the occurrence of E. Coli
- …
