Which of the following best describes the purpose of heuristic in programming?

Introduction

Heuristics are approximate strategies or ‘rules of thumb’ for decision making and problem solving that do not guarantee a correct solution but that typically yield a reasonable solution or bring one closer to hand. As such, they stand in contrast to algorithms that will produce a correct solution given complete and correct inputs. More specifically, heuristics are usually thought of as shortcuts that allow decisions or solutions to be reached more rapidly and in conditions of incomplete or uncertain information—often because they do not process all the available information. Decision heuristics have been studied in different research traditions, primarily one that has focused on when and where verbally described heuristics can break down and yield biases, that is, deviations from classical norms of rationality, and another that has investigated how specific computationally modeled heuristicscan exploit structured information to yield fast and accurate decisions. Heuristics proposed for probability judgments include representativeness, availability, and anchoring-and-adjustment; for choices between alternatives, heuristics include recognition, one-reason decision making, and cue tallying; and for sequential search across alternatives, satisficing (searching with an aspiration level) is a common heuristic approach.

Evaluating the Performance of Heuristics

Heuristics, as well as any model of decision processes, can be evaluated according to how well they describe people's choices and how well they correspond to normative criteria of accuracy. What normative criteria can be used depends on the nature of the task. In preference tasks, such as choosing between consumer products or monetary gambles, the accuracy of a heuristic can only be determined by assessing how closely its predictions match that of some gold standard for rational preferences, such as maximizing expected utility. In inference tasks, such as which soccer team will win or which of two cities is larger, the accuracy of heuristics is judged by their ability to predict real-world outcomes. The same, or similar, heuristics can be used in both preference tasks and inference tasks, but some were traditionally studied in the preference and some in the inference context. There is also a tradition of studying heuristics that are used primarily in social contexts. Success of these heuristics is also judged by criteria such as fairness, coordination, and equality.

Heuristics

Heuristics are the actual mental rules humans employ when making decisions such as choosing an action from a set of alternatives or deciding when to stop searching for a better option. A simple illustration is ‘one-reason decision making’ according to which individuals make choices based on the first relevant criterion that discriminates between the alternatives (Gigerenzer and Todd, 1999). Heuristics generally deviate from optimizing behavior in the sense that individuals are not assumed to gather all relevant information about all available alternatives before making a decision, even when such an exhaustive search was possible. Heuristics are adaptive decision-making rules (Gigerenzer and Selten, 2001) because they are in accord not only with human cognitive abilities but also with the information structure of the environment (Simon, 1956). Since they deviate from optimizing behavior, heuristics are necessarily linked to satisficing behavior (March and Simon, 1958). Originally conceptualized as ‘deviations from perfect rationality’ in the biases and heuristics program of Kahneman et al. (1982), work on heuristic decision making has shown that simple heuristics can actually outperform more advanced decision models. Recent developments in the decision literature focus on a ‘dual mode model of decision making’ distinguishing two mental systems, one fast, requiring little effort, and driven by heuristics, and one slow, effortful, and driven by conscious deliberation (Kahneman, 2011). Examining the interactions between these two systems and the conditions that determine which of the two systems is activated is a main frontier of this field of research.

2.4.5.2 Heuristics for PDPTW

Recall that exact methods usually are used for small problems (due to long computation time). Exact algorithms include dynamic programming techniques, column generation methods, branch-and-cut, branch-and-price solvers, and more [32]. Among approximate methods, the local search heuristics are commonly used to solve PDP or PDPTW.

The heuristics operators can be classified as follows [229]:

Intraroute operators, for example, 2-opt or rearrange operator.

Interroute operators, for example, an exchange or shift operator.

Ropke and Pisinger [210] used the heuristic called Adaptive Large Neighbor Search (ALNS) to solve PDPTW. It is an extended version of Large Neighbor Search (LNS), which removes and reinserts requests for a given solution to find a better solution. ALNS uses various removal and insertion heuristics. The removal heuristics use relatedness measure (the measure of similarity between two requests), worst removal measure (the cost of solution without the given request), or random selection. The insertion heuristic may use the regret heuristic, which improves the basic greedy algorithm by inserting the request at the position with the lowest cost.

Heuristics

Heuristics are cognitive shortcuts that can be applied consciously or unconsciously to reduce the complexity of decisions. Initially research emphasized the shortcomings of heuristics. The ‘heuristics-and-biases’ approach showed that people often rely on simple decision rules but that they would be better off – in terms of accuracy – if they would not. This approach goes back to the seminal work of Kahneman and Tversky. The three heuristics that received most attention were availability, representativeness, and anchoring and adjustment. The availability heuristic refers to the tendency to assess the probability of an event based on the ease with which instances of that event come to mind. (The tendency to assess the probability of an event based on the ease with which instances of that event come to mind.) Generally people overestimate the probability of an event if concrete instances of that event are easily accessible in memory. Ease of recall and frequency of occurrence are often correlated. A number of factors that affect memory are, however, unrelated to probability. For example, vivid images are easier to recall than pallid ones. Thus having been involved in a serious car accident is likely to be better remembered than annual statistics about the frequency of (types of) traffic accidents. The former is likely to have more impact on probability estimates than the latter.

The representativeness heuristic refers to the tendency to assess the probability that a stimulus belongs to a particular class by judging the degree to which that event corresponds to an appropriate mental model. A well-known example of how ignoring prior probabilities can affect judgment was reported by Kahneman and Tversky in 1973 (see Kahneman, 2011). In their study, respondents were provided with brief personality sketches, supposedly of engineers and lawyers. They were asked to assess the probability that each sketch described a member of one profession or the other. Half the respondents were told the population from which the sketches were drawn consisted of 30 engineers and 70 lawyers, the remaining respondents were told that there were 70 engineers and 30 lawyers. Findings showed that the prior probabilities were essentially ignored, and that respondents estimated the probability of class membership by judging how similar each personality sketch was to their mental model of an engineer or a lawyer.

Anchoring and adjustment refers to a general judgment process in which an initially given or generated response serves as an anchor, and other information is insufficiently used to adjust that response. The anchoring and adjustment heuristic is based on the assumption that people often start their judgmental process by focusing on some initial value that serves as an anchor. The biases related to this heuristic stem from the use of irrelevant anchors and the tendency to insufficiently adjust up or down from an original starting value or anchor.

Initially these three heuristics were described along with a dozen systematic biases such as nonregressive prediction, neglect of base-rate information, and overconfidence. Both heuristics and biases were seen as causing systematic errors in estimates of known quantities and statistical facts. Kahneman and Frederick (see Kahneman, 2011) revised the early conception of heuristics and biases and proposed a new formulation – highlighting a common process of attribute substitution – to explain how heuristics work. Moreover, their new model also includes an explicit treatment of the conditions under which intuitive judgments are modified or overrated by more elaborate cognitive processes. In this new approach, the word heuristic is used in two senses: The noun refers to the cognitive process, and the adjective in heuristic attribute specifies the attribute that is substituted in a particular judgment. Kahneman and Frederick added a family of prototype heuristics in which an attribute of a prototype is substituted for an extensional attribute of its category; the original instance of a so-called prototype heuristic is the use of representativeness in category prediction we discussed earlier. (A heuristic in which an attribute of a prototype is substituted for an extensional attribute of its category.)

Gigerenzer and colleagues stressed the adaptive role of heuristics. In their view, heuristics are efficient cognitive processes that ignore part of the information: Moreover, how well cognitive heuristics function in decisions under uncertainty is in their view an empirical question. Gigerenzer and Gaissmaier (2011) refer to simple, fast, and frugal heuristics that ‘make people smart.’ (Various classes of adaptive heuristics that people use to make decisions faster, more frugally, and/or more accurately.) They present an overview of the various heuristics people use in different types of situations, and also attempt to answer the question when people should rely on a given heuristic rather than a more complex decision strategy. Gigerenzer and colleagues stress the conscious and deliberate use of heuristics as strategies that ignore information to make decisions faster, more frugally, and/or more accurately than more complex methods. They also refer to less-is-more effects; i.e., when less information or computation leads to more accurate judgments than more information or computation. They thus disagree with the view of Payne et al. (1992) that people rely on heuristics because information search and computation costs time and effort, and that heuristics save time and effort at the expense of accuracy.

A first class of heuristics introduced by Gigerenzer and colleagues can be related to an important capacity of our memory; i.e., the fact that a sense of recognition appears in consciousness earlier than recollection. The recognition heuristic simply states that if one of two alternatives is recognized and the other is not, then one should infer that the recognized alternative has the higher value with respect to the criterion. An example is name recognition of cities; this turns out to be a reasonably valid predictor of their population. The higher the recognition validity (α) for a given criterion, the more ecologically rational it is to rely on the recognition heuristic. For each individual, α can be computed by α = C/(C + W) where C is the number of correct inferences the recognition heuristic would make, computed across all pairs in which one alternative is recognized and the other is not, and W is the number of wrong inferences.

The fluency heuristic states that if both alternatives are recognized but one is recognized faster, then infer that this alternative has the higher value with respect to the criterion. The fluency heuristic is ecologically rational if the speed of recognition is correlated with the criterion. Fluency also plays a role when alternatives are not given but need to be generated from memory. Johnson and Raab (2003) introduced the take-the-first heuristic: Choose the first alternative that comes to mind. They tested the ecological validity of this approach with experienced handball players when choosing how to proceed in a given situation (e.g., pass the ball to another player, or take a shot). A third class of heuristics are so-called one-reason decisions: a class of heuristics that bases judgments on one good reason only, ignoring other cues. Obviously this heuristic is related to the lexicographic decision rule: The latter is defined in terms of the ‘most important’ attribute. One-clever-clue heuristics (an example of one-reason decision making) focuss on the usefulness and predictive value of the selected cue of the overall outcome of the choice. Similarly, the take-the-best heuristic is a model of how people infer which of two alternatives has a higher value on a criterion, based on binary cue values retrieved from memory. Take-the-best consists of three steps: (1) search through cues in order of their validity, (2) stop on finding the first cue that discriminates between the alternatives, and (3) the alternative with the more positive cue value has the higher criterion value.

Karelaia (2006) showed that a confirmatory stopping rule – stop after two cues are found that point to the same alternative – leads to quite robust results. This heuristic tends to be ecologically valid in situations where the decision maker knows little about the validity of cues. This class of heuristics weighs cues or alternatives equally and thus makes simple trade-offs. Tallying is an example. Tallying entails simply counting the number of cues favoring one alternative in comparison to others. In three steps: (1) search through cues, (2) stopping rule: if the number of cues is the same for both alternatives, search for another cue. If no more cues are found, guess. Otherwise (3) decide for the alternative that is favored by more cues.

Basically Gigerenzer's work showed that heuristics are not good or bad and – in some circumstances – can be more accurate than more complex strategies. More research is needed on how people learn to use heuristics in an adaptive way; i.e., when to use the proper strategy from their adaptive heuristic tool box.

Which of the following best describes what the purpose of a heuristic in programming?

Why might a programmer decide to make a portion of an algorithm heuristic?

Which of the following is a consideration when determining the efficiency of an algorithm?

Which if the following best describes the ability of parallel computing solutions to improve efficiency?