A Fusion-Based Multi-Information Source Optimization Approach using Knowledge Gradient Policies

AbstractSearching for accurate machine and deep learning models is a computationally expensive and awfully energivorous process. A strategy which has been recently gaining importance to drastically reduce computational time and energy consumed is to exploit the availability of different information sources, with different computational costs and different “fidelity,” typically smaller portions of a large dataset. The multi-source optimization strategy fits into the scheme of Gaussian Process-based Bayesian Optimization. An Augmented Gaussian Process method exploiting multiple information sources (namely, AGP-MISO) is proposed. The Augmented Gaussian Process is trained using only “reliable” information among available sources. A novel acquisition function is defined according to the Augmented Gaussian Process. Computational results are reported related to the optimization of the hyperparameters of a Support Vector Machine (SVM) classifier using two sources: a large dataset—the most expensive one—and a smaller portion of it. A comparison with a traditional Bayesian Optimization approach to optimize the hyperparameters of the SVM classifier on the large dataset only is reported.

Download Full-text

Effects of information source, pedigree, and reliability on operators' utilizaton of diagnostic advice

PsycEXTRA Dataset ◽

10.1037/e577392012-057 ◽

2005 ◽

Author(s):

Poornima Madhavan ◽

Douglas A. Wiegmann

Keyword(s):

Information Source

Download Full-text

Is corporate blogging a good way to spread word-of-mouth? The role of the information source and comments of others on consumers' attitudes and word-of-mouth intentions

PsycEXTRA Dataset ◽

10.1037/e621092012-185 ◽

2009 ◽

Author(s):

Jean-Francois Belisle ◽

Sylvain Senecal ◽

Lionel Bohbot

Keyword(s):

Word Of Mouth ◽

Information Source

Download Full-text

Interactive multiobjective DEA target setting using lexicographic DDF

RAIRO - Operations Research ◽

10.1051/ro/2019105 ◽

2020 ◽

Vol 54 (6) ◽

pp. 1703-1722 ◽

Cited By ~ 1

Author(s):

Narges Soltani ◽

Sebastián Lozano

Keyword(s):

Efficient Frontier ◽

Directional Distance Function ◽

Fixed Number ◽

Optimization Approach ◽

Target Setting ◽

Interactive Process ◽

Decision Making Unit ◽

Setting Approach ◽

Promising Area ◽

Current Region

In this paper, a new interactive multiobjective target setting approach based on lexicographic directional distance function (DDF) method is proposed. Lexicographic DDF computes efficient targets along a specified directional vector. The interactive multiobjective optimization approach consists in several iteration cycles in each of which the Decision Making Unit (DMU) is presented a fixed number of efficient targets computed corresponding to different directional vectors. If the DMU finds one of them promising, the directional vectors tried in the next iteration are generated close to the promising one, thus focusing the exploration of the efficient frontier on the promising area. In any iteration the DMU may choose to finish the exploration of the current region and restart the process to probe a new region. The interactive process ends when the DMU finds its most preferred solution (MPS).

Download Full-text

Abstracting lexocigraphical works

Scientific and Technical Libraries ◽

10.33186/1027-3689-2018-6-23-36 ◽

2018 ◽

pp. 23-36

Author(s):

V. V. Goncharova

Keyword(s):

Information Processing ◽

Information Source ◽

Scientific Research ◽

Information Value ◽

International Standards ◽

Information Compression ◽

Current Trends ◽

User Groups ◽

Secondary Information ◽

Negative Factors

The increasing interest towards abstracting as a type of analytical and synthetical information processing due to science globalization trend, is emphasized. The professionals who study this primary information compression are bibliographers, linguists, and information specialists. The author argues that modern professors and students all have to and must learn abstracting in accordance with the international standards for scientific, research, reference and instructional works.The author points to the diversity of the national lexicographical studies and, based on the abstracts index obtained as a result of her study, characterizes the current trends in abstracting linguistic dictionaries. The key user groups are defined. Publishers’ abstracts of dictionaries are discussed and represented. The example of dictionary Internet-based abstract analysis is given (50 items). Based on the abstracts texts, main negative factors to impact information value of this secondary information source are revealed, that is: lacking data essential for users, incomplete description of targeted readership, etc.The author introduces a model plan for digital guides of Russian lexicographical works and complements the plan with the systematic aspect analysis. She concludes that abstracting is an intellectually intensive process. It is underexplored as far as lexicographical works are concerned, and offers many possibilities for further studies.

Download Full-text