On Stopping Evidence Gathering for Diagnostic Bayesian Networks

Gaag, Linda C. van der; Bodlaender, Hans L.

doi:10.1007/978-3-642-22152-1_15

Cited by 14 publications

(17 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…But optimizing VoI in any uncertain system has been shown to be extremely hard [14,1]. Although non-myopic (non-greedy) optimization algorithms have been proposed for restricted classes of probabilistic graphical models [13,14,7], majority of optimization approaches are myopic (greedy) [27,5]. Since the ProbLog programs we consider are not restricted to modeling particular types of graphical models, we propose an algorithm that, instead of π * , creates a plan π by greedily choosing each of its steps based on Definition 7.…”

Section: Greedy Optimization Of Voimentioning

confidence: 99%

Value of Information in Probabilistic Logic Programs

Ghosh

Ramakrishnan

2019

Electron. Proc. Theor. Comput. Sci.

View full text Add to dashboard Cite

In medical decision making, we have to choose among several expensive diagnostic tests such that the certainty about a patients health is maximized while remaining within the bounds of resources like time and money. The expected increase in certainty in the patient's condition due to performing a test is called the value of information (VoI) for that test. In general, VoI relates to acquiring additional information to improve decision-making based on probabilistic reasoning in an uncertain system. This paper presents a framework for acquiring information based on VoI in uncertain systems modeled as Probabilistic Logic Programs (PLPs). Optimal decision-making in uncertain systems modeled as PLPs have already been studied before. But, acquiring additional information to further improve the results of making the optimal decision has remained open in this context.We model decision-making in an uncertain system with a PLP and a set of top-level queries, with a set of utility measures over the distributions of these queries. The PLP is annotated with a set of atoms labeled as "observable"; in the medical diagnosis example, the observable atoms will be results of diagnostic tests. Each observable atom has an associated cost. This setting of optimally selecting observations based on VoI is more general than that considered by any prior work. Given a limited budget, optimally choosing observable atoms based on VoI is intractable in general. We give a greedy algorithm for constructing a "conditional plan" of observations: a schedule where the selection of what atom to observe next depends on earlier observations. We show that, preempting the algorithm anytime before completion provides a usable result, the result improves over time, and, in the absence of a well-defined budget, converges to the optimal solution.

show abstract

Section: Greedy Optimization Of Voimentioning

confidence: 99%

Value of Information in Probabilistic Logic Programs

Ghosh

Ramakrishnan

2019

Electron. Proc. Theor. Comput. Sci.

View full text Add to dashboard Cite

show abstract

“…These predictive processing sub-processes (belief revision and adding observations) correspond to aspects of parameter tuning and sensitivity analysis [6] and selecting evidence [38]. Algorithmic and analytical aspects of these problems are of direct relevance to the Bayesian Brain hypothesis.…”

Section: Meta-theory Of Bayesian Networkmentioning

confidence: 99%

Functional Representation of Prototypes in LVQ and Relevance Learning

Melchert¹,

Seiffert

Biehl³

2016

Advances in Self-Organizing Maps and Learning Vector Quantization

View full text Add to dashboard Cite

Held yearly, the objective of BNAIC is to promote and disseminate recent research developments in Artificial Intelligence within Belgium, Luxembourg and the Netherlands. However, it does not exclude contributions from countries outside the Benelux. As in previous years, BNAIC 2016 welcomed four types of contributions, namely A) regular papers, B) compressed contributions, C) demonstration abstracts, and D) thesis abstracts.We received 93 submissions, consisting of 24 regular papers, 47 short papers, 11 demonstration abstracts and 11 thesis abstracts. After a thorough review phase by the Program Committee, the conference chairs made the final acceptance decisions. The overall acceptance rate was 88% (63% for regular papers, 100% for compressed contributions and demonstration abstracts, and 91% for thesis abstracts).In addition to the regular research presentations, posters and demonstrations, we were happy to include several other elements in the program of BNAIC 2016, among To conclude, we want to express our gratitude to all people who made this conference possible: in addition to all invited speakers mentioned above, many thanks to all organizing and program committee members for their hard work in assuring the high quality of this conference. Moreover, we wish to thank all student volunteers, administrative and secretarial assistants, and of course our sponsors. We also gratefully acknowledge help from the BNVKI and from previous organizers. And last, but certainly not least, we cordially thank all the authors who made important contributions to the conference. Without their efforts, this conference could not have taken place. Inference that aims at determining whether a hypothesis is entailed by a text. Usually tackled by machine learning techniques employing features which represent similarity between texts, the recent availability of more training data presupposes that Neural Networks that are able to learn latent feature from data for generalized prediction could be employed. This paper employs the Child-Sum Tree-LSTM for solving the challenging problem of textual entailment. Our approach is simple and able to generalize well without excessive parameter optimization. Evaluation done on SNLI, SICK and other TE datasets shows the competitiveness of our approach.

show abstract

“…These works concentrate on the advancement of data, regularly characterized as either joint entropy or data pick up (delta entropy). It is regular practice to utilize the covetous (nearsighted) arrangement towards this determination issue [40], [38], [39], with ensured execution limits, because of the handling complexities of discovering the ideal arrangement [28], [29], [30]. SmartContext expands upon the insatiable arrangement of Krause and Guestrin [40], however is adjusted to utilizing estimation precision rather than data pick up.…”

Section: Literature Surveymentioning

confidence: 99%

Functional Context Awareness: Measuring and Utilizing the Context Dependency of Mobile Usage

2016

IJSR

View full text Add to dashboard Cite

Context data brings new open doors for productive and compelling applications and services on mobile devices. An extensive variety of examination has abused context dependency, i.e. the relations in the middle of context(s) and the result, to accomplish huge, evaluated, execution picks up for an assortment of applications and services. These works ordinarily need to manage the difficulties of numerous context sources prompting an inadequate preparing dataset, and the difficulties of energy hungry context sensors. Regularly, they address these difficulties in an application particular and specially appointed way. We free mobile application planners and analysts from these weights by giving a systematic way to deal with these difficulties. Specifically, we 1) characterize and measure the context-dependency of three central sorts of mobile use (went by sites, telephone calls, and application use) in an application freethinker yet viable way, giving understanding into the execution of potential application. 2) Address the test of information meager condition when managing various context sources in an efficient way. 3) Present SmartContext to address the energy challenge via consequently selecting among context sources while guaranteeing a base precision for every estimation. Our examination and discoveries depend on one year of utilization and context follows gathered, in actuality, settings from 24 iPhone clients. We show discoveries with respect to the context dependency of three sorts of mobile utilization from 24 clients, yet our philosophy and the lessons we learn can be promptly stretched out to different sorts of use and additionally framework assets. Our discoveries control the improvement of context mindful frameworks, and highlight the difficulties and assumptions in regards to the context dependency of mobile use.

show abstract

On Stopping Evidence Gathering for Diagnostic Bayesian Networks

Cited by 14 publications

References 10 publications

Value of Information in Probabilistic Logic Programs

Value of Information in Probabilistic Logic Programs

Functional Representation of Prototypes in LVQ and Relevance Learning

Functional Context Awareness: Measuring and Utilizing the Context Dependency of Mobile Usage

Contact Info

Product

Resources

About