Bernoulli's principle of insufficient reason and conservation of information in computer search

Abstract

Abstract-Conservation of information (COI) popularized by the no free lunch theorem is a great leveler of search algorithms, showing that on average no search outperforms any other. Yet in practice some searches appear to outperform others. In consequence, some have questioned the significance of COI to the performance of search algorithms. An underlying foundation of COI is Bernoulli's Principle of Insufficient Reason 1 (PrOIR) which imposes of a uniform distribution on a search space in the absence of all prior knowledge about the search target or the search space structure. The assumption is conserved under mapping. If the probability of finding a target in a search space is p, then the problem of finding the target in any subset of the search space is p. More generally, all some-to-many mappings of a uniform search space result in a new search space where the chance of doing better than p is 50-50. Consequently the chance of doing worse is 50-50. This result can be viewed as a confirming property of COI. To properly assess the significance of the COI for search, one must completely identify the precise sources of information that affect search performance. This discussion leads to resolution of the seeming conflict between COI and the observation that some search algorithms perform well on a large class of problems.Index Terms-active information, Bernoulli's principle of insufficient reason, Bernoulli's principle of non-sufficient reason, Bertrand's paradox, conservation of generalization performance, conservation of information, endogenous information, equal distribution of ignorance, evolutionary search, no free lunch theorem, principle of indifference, uniform distribution

Triple Bottomline Many-Objective-Based Decision Making for a Land Use Management Problem

“…Given the uncertainties around the evaluation of the objectives, we incorporated them in the HRV in a way that enhanced visual steering, allowing a more informed and responsible decision‐making process. This was achieved by using uncertainty ranges based on a concept borrowed from the Physical Programming approach for multi‐objective optimization (Dembski and Marks, ). A percent confidence/deviation value for each objective attribute in the trade‐off set made it possible to compute the corresponding range bounds for each trade‐off point: where i is theobjective index, j is the trade‐off point index and δ i is the allowable percent deviation value for each objective = [3, 10, 10, 10, 3, 1, 1, 1, 2, 5, 5, 2, 5, 10] for water quantity, nitrate leaching, phosphorus loss, CO 2 e, sediments, discounted costs, discounted income, discounted EBIT, milksolids, sawlog, pulpwood, beef, sheepmeat and wool, respectively.…”

“…Given the uncertainties around the evaluation of the objectives, we incorporated them in the HRV in a way that enhanced visual steering, allowing a more informed and responsible decision-making process. This was achieved by using uncertainty ranges based on a concept borrowed from the Physical Programming approach for multi-objective optimization (Dembski and Marks, 2009). A percent confidence/deviation value for each objective attribute in the trade-off set made it possible to compute the corresponding range bounds for each trade-off point:…”

Information transmission through genetic algorithm fitness maps

“…2) Fitness Value Representations: If the distribution of fitness values in a fitness map is known beforehand, a compression scheme such as Huffman encoding can be used to assign short binary strings to frequently occurring fitness values and longer strings to less frequently occurring values. If distribution information is not available, however, we assume an equiprobable distribution of values, so that each value in the set Y of possible fitness values has a 1 |Y| probability of occurring [19], [20], [21]. This assumption results in an encoding length of l = log 2 |Y| binary digits per value, following Shannon's entropy formula [11].…”

Conservation of information in relative search performance

“…As illustrated in Figure 1, assume Pirate X searches for the location in a specific order such as (1,2,3). Assuming uniformity [3], the probability of the treasure being found in any of those locations is However, this changes if both pirates are hunting for treasure at the same time.…”

“…Assuming uniformity [3], the probability of the treasure being found in any of those locations is However, this changes if both pirates are hunting for treasure at the same time. Pirate Y has chosen locations such that in two of the three cases, he will have searched a location and taken the treasure before Pirate X.…”