Importance sampling is a general technique for estimating the properties of a particular distribution, while only having samples generated from a different distribution rather than the distribution of interest. Depending on the application, the term may refer to the process of sampling from this alternative distribution, the process of inference, or both.
More formally, let X be a random variable in S. A random variable is a rigorously defined mathematical entity used mainly to describe Chance and Probability in a mathematical way Let p be a probability measure on S, and f some function on S. A probability space, in Probability theory, is the conventional Mathematical model of Randomness. Then the expectation of f under p can be written as
If we have random samples , generated according to p, then an empirical estimate of p is
where for xi = x and 0 otherwise.
In that case, we can easily obtain the Monte-Carlo empirical estimate of
Unfortunately, when the samples are generated from a different distribution than the one that we are interested in, we can no longer use this straightforward estimate. However, we may use the importance sampling technique, which consists of placing different importance on each sample, depending on how likely it was for it to have been generated by the distribution that we're interested in, p, rather than the actual sampling distribution, q.
More formally, consider another probability measure, q, with the same support as p. In Mathematics, the support of a function is the set of points where the function is not zero or the closure of that set From the definition of the expectation given above, we have
where , is known as the importance weight and the distribution q is frequently referred to as the sampling or proposal distribution. Then, if we have random samples , generated according to q, a Monte Carlo estimate of follows from the above equation by viewing the problem as that of estimating the expectations and .
where are the normalised importance weights.
The technique is completely general and the above analysis can be repeated essentially exactly also for other choices of p, for example when it represents a conditional distribution. Note that when p is the uniform distribution, we are just estimating the (scaled) integral of f over S, so the method can also be used for estimating simple integrals.
There are two main applications of importance sampling methods, which naturally, are interrelated. While the aim of both applications is to estimate statistics of random variables, the field of probabilistic inference focuses more on the estimation of p or related statistics, while the field of simulation focuses more on the choice of the distribution q. Nevertheless, the basic theory and tools are identical.
Such methods are frequently used to estimate posterior densities or expectations in state and/or parameter estimation problems in probabilistic models that are too hard to treat analytically, for example in Bayesian networks. A Bayesian network (or a belief network) is a Probabilistic graphical model that represents a set of Variables and their probabilistic independencies
Importance sampling (IS) is a variance reduction technique that can be used in the Monte Carlo method. In mathematics more specifically in the theory of Monte Carlo methods variance reduction is a procedure used to increase the precision of the estimates that can be obtained Monte Carlo methods are a class of Computational Algorithms that rely on repeated Random sampling to compute their results The idea behind IS is that certain values of the input random variables in a simulation have more impact on the parameter being estimated than others. A random variable is a rigorously defined mathematical entity used mainly to describe Chance and Probability in a mathematical way Simulation is the imitation of some real thing state of affairs or process If these "important" values are emphasized by sampling more frequently, then the estimator variance can be reduced. In Statistics, an estimator is a function of the observable sample data that is used to estimate an unknown population Parameter (which is called the Hence, the basic methodology in IS is to choose a distribution which "encourages" the important values. This use of "biased" distributions will result in a biased estimator if it is applied directly in the simulation. However, the simulation outputs are weighted to correct for the use of the biased distribution, and this ensures that the new IS estimator is unbiased. The weight is given by the likelihood ratio, that is, the Radon-Nikodym derivative of the true underlying distribution with respect to the biased simulation distribution. The likelihood ratio often denoted by \Lambda (the capital Greek letter Lambda) is the ratio of the maximum Probability of a Result In Mathematics, the Radon–Nikodym theorem is a result in Functional analysis that states that given a measurable space ( X,&Sigma if a
The fundamental issue in implementing IS simulation is the choice of the biased distribution which encourages the important regions of the input variables. Choosing or designing a good biased distribution is the "art" of IS. The rewards for a good distribution can be huge run-time savings; the penalty for a bad distribution can be longer run times than for a general Monte Carlo simulation without importance sampling.
Consider estimating by simulation the probability of an event , where X is a random variable with distribution F and probability density function , where prime denotes derivative. In Probability theory and Statistics, a probability distribution identifies either the probability of each value of an unidentified Random variable In Mathematics, a probability density function (pdf is a function that represents a Probability distribution in terms of Integrals Formally a probability In Calculus, a branch of mathematics the derivative is a measurement of how a function changes when the values of its inputs change A K-length independent and identically distributed (i. "IID" or "iid" redirects here For other uses see IID (disambiguation. i. d. ) sequence is generated from the distribution F, and the number kt of random variables that lie above the threshold t are counted. The random variable kt is characterized by the Binomial distribution
Importance sampling is concerned with the determination and use of an alternate density function (for X), usually referred to as a biasing density, for the simulation experiment. WikipediaWikiProject Probability#Standards for a discussion of standards used for probability distribution articles such as this one This density allows the event to occur more frequently, so the sequence lengths K gets smaller for a given estimator variance. In Statistics, an estimator is a function of the observable sample data that is used to estimate an unknown population Parameter (which is called the Alternatively, for a given K, use of the biasing density results in a variance smaller than that of the conventional Monte Carlo estimate. From the definition of , we can introduce as below.
is a likelihood ratio and is referred to as the weighting function. The last equality in the above equation motivates the estimator
This is the IS estimator of and is unbiased. That is, the estimation procedure is to generate i. i. d. samples from and for each sample which exceeds , the estimate is incremented by the weight evaluated at the sample value. The results are averaged over trials. The variance of the IS estimator is easily shown to be
Now, the IS problem then focuses on finding a biasing density such that the variance of the IS estimator is less than the variance of the general Monte Carlo estimate. For some biasing density function, which minimizes the variance, and under certain conditions reduces it to zero, it is called an optimal biasing density function.
Although there are many kinds of biasing methods, the following two methods are most widely used in the applications of IS.
Shifting probability mass into the event region by positive scaling of the random variable with a number greater than unity has the effect of increasing the variance (mean also) of the density function. This results in a heavier tail of the density, leading to an increase in the event probability. Scaling is probably one of the earliest biasing methods known and has been extensively used in practice. It is simple to implement and usually provides conservative simulation gains as compared to other methods.
In IS by scaling, the simulation density is chosen as the density function of the scaled random variable , where usually a > 1 for tail probability estimation. By transformation,
and the weighting function is
While scaling shifts probability mass into the desired event region, it also pushes mass into the complementary region which is undesirable. If is a sum of random variables, the spreading of mass takes place in an dimensional space. The consequence of this is a decreasing IS gain for increasing , and is called the dimensionality effect.
Another simple and effective biasing technique employs translation of the density function (and hence random variable) to place much of its probability mass in the rare event region. Translation does not suffer from a dimensionality effect and has been successfully used in several applications relating to simulation of digital communication systems. Digital communications refers to the transmission of a sequence of Digital messages (a Bit stream) or a digitized analog signal It often provides better simulation gains than scaling. In biasing by translation, the simulation density is given by
where is the amount of shift and is to be chosen to minimize the variance of the IS estimator.
The fundamental problem with IS is that designing good biased distributions becomes more complicated as the system complexity increases. Complex systems are the systems with long memory since complex processing of a few inputs is much easier to handle. This dimensionality or memory can cause problems in three ways:
In principle, the IS ideas remain the same in these situations, but the design becomes much harder. In Telecommunication, intersymbol interference ( ISI) is a form of Distortion of a signal in which one symbol interferes with A viterbi decoder uses the Viterbi algorithm for decoding a bitstream that has beenencoded using Forward error correction based on a Convolutional code A successful approach to combat this problem is essentially breaking down a simulation into several smaller, more sharply defined subproblems. Then IS strategies are used to target each of the simpler subproblems. Examples of techniques to break the simulation down are conditioning and error-event simulation (EES) and regenerative simulation.
In order to identify successful IS techniques, it is useful to be able to quantify the run-time savings due to the use of the IS approach. The performance measure commonly used is , and this can be interpreted as the speed-up factor by which the IS estimator achieves the same precision as the MC estimator. This has to be computed empirically since the estimator variances are not likely to be analytically possible when their mean is intractable. Other useful concepts in quantifying an IS estimator are the variance bounds and the notion of asymptotic efficiency.
Variance is not the only possible cost function for a simulation, and other cost functions, such as the mean absolute deviation, are used in various statistical applications. In Mathematics, the term optimization, or mathematical programming, refers to the study of problems in which one seeks to minimize or maximize a real function Nevertheless, the variance is the primary cost function addressed in the literature, probably due to the use of variances in confidence intervals and in the performance measure . In Statistics, a confidence interval (CI is an interval estimate of a Population parameter.
An associated issue is the fact that the ratio overestimates the run-time savings due to IS since it does not include the extra computing time required to compute the weight function. Hence, some people evaluate the net run-time improvement by various means. Perhaps a more serious overhead to IS is the time taken to devise and program the technique and analytically derive the desired weight function.