E. D. Solojentsev, V. V. Karassev. I³-technologies for management of risk in economics
The problem of management of the non-success risk at decision of difficult economic
problems is formulated. We state the basic theses of I³-technologies with risk
LP-models and knowledge
bases (KB). Classes of risk LP-models and procedures of I³-technologies
are described. We state applications of I³-technologies for
management of risk in different areas. We state I³-technologies and
risk LP-models for identification of bribes and corruption by statistical data.
We prove it is impossibility efficiently to decide the difficult problem of
counteraction to bribes and corruption in Russia without I³-technology, scientists
and public opinion.
E. D.
Solojentsev. Logic and probabilistic management of invalidity risk of debugging
tests of machines, processes and systems
We state the
basis of logical-probabilistic (LP) risk management by invalid debugging tests
(DT) of machines, processes and systems. We use the example of the
meaningful engineering descriptions of scenarios of failures and breakdowns in
DT of engineering system. We present LP-descriptions of events at DT,
introduce events and the logical Boolean variables for initiating,
derivatives, and the outcome of events. We offer the risk L-model of
invalid DT as a disjunction of the shortest paths (conjunctions) of derivative
invalid events. Risk L-function has repeated elements. This complicates
the calculations of the risk P-function. We investigate the risk
LP-model of invalid DT of the technical system, identify the
structural importance of initiating events and their contributions to the risk
of invalid DT. We develop the technique of risk management of first
invalid DT and the scheme of risk management the whole process of
invalid DT of several stages. We state features of risk management of
operational and evolution invalid tests.
V. V. Alexeev, V.V Karassev, E. D. Solojentsev. Training and testing of
risk LP-model by statistical data
We consider the
problem of testing risk LP-models as the problem of forecasting by the
risk LP-model. We establish wrong of approval on the
uselessness of testing for assessment of accuracy
of risk LP-model. We investigate the influence of various factors on the
accuracy of training and testing (forecasting) of logical-probabilistic (LP)
risk models. We find practical impossibility of generating the
identical training and testing samples for the risk LP-model due to
the large number of events-gradations, of their different
frequencies in the samples and their varying significance in the risk. We propose
the method of asymmetric training of risk LP-model to improve the accuracy of forecasting.
In this method ratio of errors of classification of good
and bad objects equal to 1 or to given value. We formulate
recommendations for training, testing and using risk LP-models.
The
logic-probabilistic (LP) model of non-success risk is proposed for decision of
difficult economic problems. The subjects that decide the problem (the State,
Banks, Scientists and Public Opinion) and the tasks to be decided are logically
connected as events. It is shown that in Russia to solve effectively
difficult economic problems without attraction of the scientists, new
technologies and public opinion is impossible. An example of non-success risk
LP-model of solving the problem of counteraction to bribes and corruption is
given.
