Exemplar-based Strategies Research Articles

Exemplar learners and rule learners: Stable tendencies or malleable preferences?

When learning new concepts, students tend to use either exemplar-based learning strategies (e.g., memorizing specific examples) or rule-based learning strategies (e.g., abstracting general rules). Prior research suggests that participants' strategy choices during learning depend on individuals' preexisting learning tendencies, with some people being exemplar learners and others rule learners. Yet, strategy choices are also influenced by how the study materials are taught (rule-focused or exemplar-focused). The present study examined how these two factors interact using an alphanumeric symbol addition task. We examined whether exemplar learners would switch to using rule-based strategies when given rule-focused training and if rule learners would fail to learn the rule when given exemplar-focused training. We found that both rule and exemplar learners used a rule-based strategy after a rule-focused training and neither group learned the rule after an exemplar-focused training. Our results suggest that individuals can be shaped to adopt either rule-based or exemplar-based strategies during learning, regardless of their inherent learning tendencies. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Zero-exemplar deep continual learning for crop disease recognition: a study of total variation attention regularization in vision transformers.

With the increasing integration of AI technology in the food industry, deep learning has demonstrated its immense potential in the domain of plant disease image recognition. However, there remains a gap in research between models capable of continual learning of new diseases and addressing the inherent catastrophic forgetting issue in neural networks. This study aims to comprehensively evaluate various learning strategies based on advanced computer vision models for multi-disease continual learning tasks in food disease recognition. To cater to the benchmark dataset requirements, we collected the PlantDiseaseCL dataset, sourced from the internet, encompassing diverse crop diseases from apples, corn, and more. Utilizing the Vision Transformer (ViT) model, we established a plant disease image recognition classifier, which, in joint learning, outperformed several comparative CNN architectures in accuracy (0.9538), precision (0.9532), recall (0.9528), and F1 score (0.9560). To further harness the potential of ViT in food disease defect recognition, we introduced a mathematical paradigm for crop disease recognition continual learning. For the first time, we proposed a novel ViT-TV architecture in the multi-disease image recognition scenario, incorporating a Total Variation (TV) distance-based loss (TV-Loss) to quantify the disparity between current and previous attention distributions, fostering attention consistency and mitigating the catastrophic forgetting inherent in ViT without prior task samples. In the incremental learning of the PlantDiseaseCL dataset across 3-Steps and 5-Steps, our strategy achieved average accuracies of 0.7077 and 0.5661, respectively, surpassing all compared Zero-Exemplar Approaches like LUCIR, SI, MAS, and even outperforming exemplar-based strategies like EEIL and ICaRL. In conclusion, the ViT-TV approach offers robust support for the long-term intelligent development of the agricultural and food industry, especially showcasing significant applicability in continual learning for crop disease image recognition.

Individual and age differences in block-by-block dynamics of category learning strategies.

The present experiment examined individual and age differences in the dynamics of category learning strategies. Participants learned categories determined by a disjunctive rule with relational features through a feedback training procedure. During training, participants responded to strategy probes following each block to provide online assessment of the extent to which rule- and exemplar-based strategies were used throughout the training period. We introduced this measure as an alternative to model-based approaches to assessing individual differences in strategy use during training. Following training, participants classified ambiguous transfer objects that were assumed to distinguish between earlier use of rule- and exemplar-based learning strategies. We included this measure to obtain a relatively objective index of strategy use during training. Next, participants provided global ratings of their use of rule- and exemplar-based strategies during training. Results showed that strategy preferences expressed on the final training block predicted categorisation of ambiguous transfer objects better than global strategy reports. In addition, we utilised the block-by-block strategy reports to investigate the dynamics of learners' strategy preferences over the course of training. The findings revealed greater fluidity in strategy preferences for both younger and older adults than has been previously documented in the category learning literature. The novel block-by-block strategy reports in conjunction with the transfer-based approach allowed for a more nuanced examination of individual and age differences in strategy use and categorisation performance.

Nothing compares: Unraveling learning task effects in judgment and categorization.

In judgment and categorization, the task is to infer the criterion value of an object based on cues. The cognitive mechanisms underlying such inferences are often distinguished in terms of whether they rely on an abstracted cue-criterion rule or on retrieving exemplars. The use of cue-based and exemplar-based strategies (and the associated generalization ability) has been shown to be influenced by how people had previously learned about the structure of the environment: learning by comparing two objects on the criterion promotes the subsequent adoption of a cue-based strategy (and generalization ability), directly learning the continuous criterion for each individual object promotes the use of an exemplar-based strategy. It is currently unclear, however, how to explain these learning task effects theoretically. We disentangled the role of three theoretically relevant differences between the two types of learning task: (a) the opportunity to directly compare the cue profiles of the objects, (b) the provision of continuous criterion information, and (c) the relative (as opposed to absolute) nature of the feedback given. In three experiments, and consistently across a categorization task and a judgment task, we found that the learning task effects seem to be driven by the relative nature of the feedback in learning by comparison, and, to a lesser extent, by the provision of the object's continuous criterion values in direct criterion learning. Cue comparison during training did not seem to contribute to the learning task effects. Our results extend existing theoretical frameworks for the use of cue-based and exemplar-based strategies. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Retrieval from memory and cue complexity both trigger exemplar-based processes in judgment

ABSTRACTJudgments and decisions can be achieved by a rule-based integration of cues or by retrieving similar exemplars from memory. In judgments based on currently available object descriptions (on-line judgments), people tend to rely on rules unless the cue-criterion relations are difficult to extract. In memory-based decisions, more evidence for exemplarbased reasoning has been found, but hitherto restricted to conditions with unknown cue polarity. These conclusions are threatened by methodological confounds concerning materials, training regime, and type of judgment task. In a factorial experiment, memory-based vs. on-line judgments as well as obvious vs. unknown cue polarity conditions were pitted against each other. Participants judged disease severity based on symptoms. Using the measurement model RulEx-J, we confirmed that both memory retrieval and unknown cue polarity have additive effects on the use of exemplar-based processes. This corroborates the idea that exemplar-based strategies are used as a backup if cue abstraction is unfeasible.

Tracing the path of forgetting in rule abstraction and exemplar retrieval.

People often forget acquired knowledge over time such as names of former classmates. Which knowledge people can access, however, may modify the judgement process and affect judgement accuracy. Specifically, we hypothesised that judgements based on retrieving past exemplars from long-term memory may be more vulnerable to forgetting than remembering rules that relate the cues to the criterion. Experiment 1 systematically tracked the individual course of forgetting from initial learning to later tests (immediate, 1 day, and 1 week) in a linear judgement task facilitating rule-based strategies and a multiplicative judgement task facilitating exemplar-based strategies. Practising the acquired judgement strategy in repeated tests helped participants to consistently apply the learnt judgement strategy and retain a high judgement accuracy even after a week. Yet, whereas a long retention interval did not affect judgements in the linear task, a long retention interval impaired judgements in the multiplicative task. If practice was restricted as in Experiment 2, judgement accuracy suffered in both tasks. In addition, after a week without practice, participants tried to reconstruct their judgements by applying rules in the multiplicative task. These results emphasise that the extent to which decision makers can still retrieve previously learned knowledge limits their ability to make accurate judgements and that the preferred strategies change over time if the opportunity for practice is limited.

Towards a Taxonomy for Classifying Intuitive Usability Prediction Strategies in Non-Experts

In order to compare the strategies consumers use to make predictions about the ease-of-use of products, the authors conducted a card sorting exercise with subject-matter experts (SMEs) to refine a coding scheme from current, widely-used usability definitions and measurement tools. Nine SMEs sorted 83 cards containing aspects found in usability definitions, guidelines, measurement tools and descriptions. The resulting fifteen clusters define a coding scheme that is both comprehensive and flexible. The refined taxonomy does not seek to redefine usability, but to provide a mechanism for researchers wanting to use a coding scheme that includes methods for classifying strategies such as usability attribution to system features, metacognitive behaviors (planning and evaluating), and exemplar-based strategies (memory for performing similar tasks).

Costs and benefits of automatization in category learning of ill-defined rules

Learning ill-defined categories (such as the structure of Medin & Schaffer, 1978) involves multiple learning systems and different corresponding category representations, which are difficult to detect. Application of latent Markov analysis allows detection and investigation of such multiple latent category representations in a statistically robust way, isolating low performers and quantifying shifts between latent strategies. We reanalyzed data from three experiments presented in Johansen and Palmeri (2002), which comprised prolonged training of ill-defined categories, with the aim of studying the changing interactions between underlying learning systems. Our results broadly confirm the original conclusion that, in most participants, learning involved a shift from a rule-based to an exemplar-based strategy. Separate analyses of latent strategies revealed that (a) shifts from a rule-based to an exemplar-based strategy resulted in an initial decrease of speed and an increase of accuracy; (b) exemplar-based strategies followed a power law of learning, indicating automatization once an exemplar-based strategy was used; (c) rule-based strategies changed from using pure rules to rules-plus-exceptions, which appeared as a dual processes as indicated by the accuracy and response-time profiles. Results suggest an additional pathway of learning ill-defined categories, namely involving a shift from a simple rule to a complex rule after which this complex rule is automatized as an exemplar-based strategy.

Cue integration vs. exemplar-based reasoning in multi-attribute decisions from memory: A matter of cue representation

AbstractInferences about target variables can be achieved by deliberate integration of probabilistic cues or by retrieving similar cue-patterns (exemplars) from memory. In tasks with cue information presented in on-screen displays, rule-based strategies tend to dominate unless the abstraction of cue-target relations is unfeasible. This dominance has also been demonstrated — surprisingly — in experiments that demanded the retrieval of cue values from memory (M. Persson & J. Rieskamp, 2009). In three modified replications involving a fictitious disease, binary cue values were represented either by alternative symptoms (e.g.,fevervs.hypothermia) or by symptom presence vs. absence (e.g.,fevervs.no fever). The former representation might hinder cue abstraction. The cues were predictive of the severity of the disease, and participants had to infer in each trial who of two patients was sicker. Both experiments replicated the rule-dominance with present-absent cues but yielded higher percentages of exemplar-based strategies with alternative cues. The experiments demonstrate that a change in cue representation may induce a dramatic shift from rule-based to exemplar-based reasoning in formally identical tasks.

Feature-Based versus Exemplar-Based Strategies in Preschoolers′ Category Learning

Five experiments examined whether preschool children learn categories by formulating a categorization rule based on attention to one or two features, or by memorizing the category assignment for whole exemplars. Using fictitious animals as stimuli, children were taught a two-category discrimination at the basic level and were tested for their category learning using a procedure that required them to categorize from memory, rather than perform perceptual matching. Transfer stimuli manipulated the retention of features encountered in the training instances, and children were instructed to classify only the items about whose category membership they were certain. By measuring the extent of generalization, in addition to categorization accuracy, it was possible to assess whether children were applying a feature-based rule in making their categorization decisions. The studies found little evidence that children were employing rules. Analyses of the individual response patterns identified few children who categorized by attending to a single attribute such as shape, or else who used a two-feature disjunctive rule. In addition, a recognition memory test given after category training produced strong evidence that children had encoded the training examples as wholes, rather than focusing on individual features. Despite clear evidence that children had encoded the training examples, however, exemplar information was not used successfully in categorization when subjects were trained on only a single instance per category. The implications of these results for theories of early categorization are discussed.