Recent Research In Artificial Intelligence Research Articles

Evolutionary Reinforcement Learning: Hybrid Approach for Safety-Informed Fault-Tolerant Flight Control

Recent research in artificial intelligence potentially provides solutions to the challenging problem of fault-tolerant and robust flight control. This paper proposes a novel Safety-Informed Evolutionary Reinforcement Learning algorithm (SERL), which combines Deep Reinforcement Learning (DRL) and neuroevolution to optimize a population of nonlinear control policies. Using SERL, the work has trained agents to provide attitude tracking on a high-fidelity nonlinear fixed-wing aircraft model. Compared to a state-of-the-art DRL solution, SERL achieves better tracking performance in nine out of ten cases, remaining robust against faults and changes in flight conditions, while providing smoother action signals.

Artificial intelligence research in hospitality: a state-of-the-art review and future directions

Purpose The purpose of this study is to analyze state-of-the-art knowledge of artificial intelligence (AI) research in hospitality. Design/methodology/approach This study adopts the theory-context-methods framework to systematically review 100 AI-related articles recently published (i.e. from 2021 to April 2023) in three top-tier hospitality journals, namely, the International Journal of Contemporary Hospitality Management, International Journal of Hospitality Management and Journal of Hospitality Marketing and Management. Findings Findings suggest that studies of AI applications in hospitality are mostly theory-driven, whereas most AI methods research adopts a data-driven approach. State-of-the-art AI applications research exhibits the most interest in service robots. In AI methods research, little attention was paid to the amid-service/experience. Research limitations/implications This study reveals inadequacies in theory, context and methods in contemporary AI research. More research from hospitality suppliers’ perspectives and research on generative AI applications are advocated in response to the unveiled research gaps and recent AI developments. Originality/value This study classifies the most recent AI research in hospitality into two main streams – AI applications research and AI methods research – and discusses the gaps in each research stream and latest AI developments. The paper then suggests future research directions to guide researchers in advancing AI research in hospitality.

Artificial intelligence in endourology: emerging technology for individualized care.

Artificial intelligence in medicine has allowed for efficient processing of large datasets to perform cognitive tasks that facilitate clinical decision-making, and it is an emerging area of research. This review aims to highlight the most pertinent and recent research in artificial intelligence in endourology, where it has been used to optimize stone diagnosis, support decision-making regarding management, predict stone recurrence, and provide new tools for bioinformatics research within endourology. Artificial neural networks (ANN) and machine learning approaches have demonstrated high accuracy in predicting stone diagnoses, stone composition, and outcomes of spontaneous stone passage, shockwave lithotripsy (SWL), or percutaneous nephrolithotomy (PCNL); some of these models outperform more traditional predictive models and existing nomograms. In addition, these approaches have been used to predict stone recurrence, quality of life scores, and provide novel methods of mining the electronic medical record for research. Artificial intelligence can be used to enhance existing approaches to stone diagnosis, management, and prevention to provide a more individualized approach to endourologic care. Moreover, it may support an emerging area of bioinformatics research within endourology. However, despite high accuracy, many of the published algorithms lack external validity and require further study before they are more widely adopted.

Automatic Windthrow Detection Using Very-High-Resolution Satellite Imagery and Deep Learning

Wind disturbances are significant phenomena in forest spatial structure and succession dynamics. They cause changes in biodiversity, impact on forest ecosystems at different spatial scales, and have a strong influence on economics and human beings. The reliable recognition and mapping of windthrow areas are of high importance from the perspective of forest management and nature conservation. Recent research in artificial intelligence and computer vision has demonstrated the incredible potential of neural networks in addressing image classification problems. The most efficient algorithms are based on artificial neural networks of nested and complex architecture (e.g., convolutional neural networks (CNNs)), which are usually referred to by a common term—deep learning. Deep learning provides powerful algorithms for the precise segmentation of remote sensing data. We developed an algorithm based on a U-Net-like CNN, which was trained to recognize windthrow areas in Kunashir Island, Russia. We used satellite imagery of very-high spatial resolution (0.5 m/pixel) as source data. We performed a grid search among 216 parameter combinations defining different U-Net-like architectures. The best parameter combination allowed us to achieve an overall accuracy for recognition of windthrow sites of up to 94% for forested landscapes by coniferous and mixed coniferous forests. We found that the false-positive decisions of our algorithm correspond to either seashore logs, which may look similar to fallen tree trunks, or leafless forest stands. While the former can be rectified by applying a forest mask, the latter requires the usage of additional information, which is not always provided by satellite imagery.

A distributional code for value in dopamine-based reinforcement learning.

Since its introduction, the reward prediction error theory of dopamine has explained a wealth of empirical phenomena, providing a unifying framework for understanding the representation of reward and value in the brain1-3. According to the now canonical theory, reward predictions are represented as a single scalar quantity, which supports learning about the expectation, or mean, of stochastic outcomes. Here we propose an account of dopamine-based reinforcement learning inspired by recent artificial intelligence research on distributional reinforcement learning4-6. We hypothesized that the brain represents possible future rewards not as a single mean, but instead as a probability distribution, effectively representing multiple future outcomes simultaneously and in parallel. This idea implies a set of empirical predictions, which we tested using single-unit recordings from mouse ventral tegmental area. Our findings provide strong evidence for a neural realization of distributional reinforcement learning.

Multitask Learning for Cross-Domain Image Captioning

Recent artificial intelligence research has witnessed great interest in automatically generating text descriptions of images, which are known as the image captioning task. Remarkable success has been achieved on domains where a large number of paired data in multimedia are available. Nevertheless, annotating sufficient data is labor-intensive and time-consuming, establishing significant barriers for adapting the image captioning systems to new domains. In this study, we introduc a novel Multitask Learning Algorithm for cross-Domain Image Captioning (MLADIC). MLADIC is a multitask system that simultaneously optimizes two coupled objectives via a dual learning mechanism: image captioning and text-to-image synthesis, with the hope that by leveraging the correlation of the two dual tasks, we are able to enhance the image captioning performance in the target domain. Concretely, the image captioning task is trained with an encoder–decoder model (i.e., CNN-LSTM) to generate textual descriptions of the input images. The image synthesis task employs the conditional generative adversarial network (C-GAN) to synthesize plausible images based on text descriptions. In C-GAN, a generative model $G$ synthesizes plausible images given text descriptions, and a discriminative model $D$ tries to distinguish the images in training data from the generated images by $G$ . The adversarial process can eventually guide $G$ to generate plausible and high-quality images. To bridge the gap between different domains, a two-step strategy is adopted in order to transfer knowledge from the source domains to the target domains. First, we pre-train the model to learn the alignment between the neural representations of images and that of text data with the sufficient labeled source domain data. Second, we fine-tune the learned model by leveraging the limited image–text pairs and unpaired data in the target domain. We conduct extensive experiments to evaluate the performance of MLADIC by using the MSCOCO as the source domain data, and using Flickr30k and Oxford-102 as the target domain data. The results demonstrate that MLADIC achieves substantially better performance than the strong competitors for the cross-domain image captioning task.

Mental labour.

Mental effort is an elementary notion in our folk psychology and a familiar fixture in everyday introspective experience. However, as an object of scientific study, mental effort has remained rather elusive. Cognitive psychology has provided some tools for understanding how effort impacts performance, by linking effort with cognitive control function. What has remained less clear are the principles that govern the allocation of mental effort. Under what circumstances do people choose to invest mental effort, and when do they decline to do so? And what regulates the intensity of mental effort when it is applied? In new and promising work, these questions are being approached with the tools of behavioural economics. Though still in its infancy, this economic approach to mental effort research has already uncovered important aspects of effort-based decision-making, and points clearly to future lines of inquiry, including some intriguing opportunities presented by recent artificial intelligence research.

A fuzzy cognitive map model to calculate a user's desirability based on personality in e-learning environments

The recent research in artificial intelligence shows an increasing interest in the modeling of human behavior factors such as personality, mood, and emotion for developing human-friendly systems. That is why there is an interest in developing models and algorithms to determine a human's emotions while interacting with a system to improve the quality of the interaction. In this paper, we propose a computational model to calculate a user's desirability based on personality in e-learning environments. The desirability is one of the most important variables in determining a user's emotions. The model receives several e-learning environmental events and predicts the desirability of the events based on the user's personality and his/her goals. The proposed model has been evaluated in a simulated and real e-learning environment. The results show that the model formulates the relationship between personality and emotions with high accuracy.

Applying Recent Argumentation Methods to Some Ancient Examples of Plausible Reasoning

Plausible (eikotic) reasoning known from ancient Greek (late Academic) skeptical philosophy is shown to be a clear notion that can be analyzed by argumentation methods, and that is important for argumentation studies. It is shown how there is a continuous thread running from the Sophists to the skeptical philosopher Carneades, through remarks of Locke and Bentham on the subject, to recent research in artificial intelligence. Eleven characteristics of plausible reasoning are specified by analyzing key examples of it recognized as important in ancient Greek skeptical philosophy using an artificial intelligence model called the Carneades Argumentation System (CAS). By applying CAS to ancient examples it is shown how plausible reasoning is especially useful for gaining a better understanding of evidential reasoning in law, and argued that it can also be applied to everyday argumentation. Our analysis of the snake and rope example of Carneades is also used to point out some ways CAS needs to be extended if it is to more fully model the views of this ancient philosopher on argumentation.

Extension of the temporal synchrony approach to dynamic variable bindingin a connectionist inference system

The relationship between symbolism and connectionism has been one of the major issues in recent artificial intelligence research. An increasing number of researchers from each side have tried to adopt the desirable characteristics of the approach. A major open question in this field is the extent to which a connectionist architecture can accommodate basic concepts of symbolic inference, such as a dynamic variable binding mechanism and a rule and fact encoding mechanism involving nary predicates. One of the current leaders in this area is the connectionist rule-based system proposed by Shastri and Ajjanagadde. The paper demonstrates that the mechanism for variable binding which they advocate is fundamentally limited, and it shows how a reinterpretation of the primitive components and corresponding modifications of their system can extend the range of inference which can be supported. Our extension hinges on the basic structural modification of the network components and further modifications of the rule and fact encoding mechanism. These modifications allow the extended model to have more expressive power in dealing with symbolic knowledge as in the unification of terms across many groups of unifying arguments.

Computational research on interaction and agency

Recent research in artificial intelligence has developed computational theories of agents' involvements in their environments. Although inspired by a great diversity of formalisms and architectures, these research projects are unified by a common concern: using principled characterizations of agents' interactions with their environments to guide analysis of living agents and design of artificial ones. This article offers a conceptual framework for such theories, surveys several other fields of research that hold the potential for dialogue with these new computational projects, and summarizes the principal contributions of the articles in this special double volume. It also briefly describes a case study in these ideas—a computer program called Toast that acts as a short-order breakfast cook. Because its designers have discovered useful structures in the world it inhabits, Toast can employ an extremely simple mechanism to decide what to do next.

Developing knowledge structures: A comparison of a qualitative-response model and two machine-learning algorithms

Many unstructured decision-making problems are too complex to be solved by existing analytical methods. Organizations commonly rely on human expertise to solve such problems. Human expertise is presumed to utilize specialized knowledge structures, and recent research in artificial intelligence has sought to devise machine-learning procedures that can derive knowledge structures from sets of data that represent problem-domain exemplars. Concurrently, research in statistical methods has resulted in the development of qualitative-response models that can be applied to some of the same kinds of problems addressed by machine-learning models-especially those that involve a classification decision. Using a difficult audit decision problem requiring expertise, this paper reports on a study comparing performance of two machine-learning algorithms and logistic regression, a widely-used qualitative-response method for classification analysis. A recently-developed machine-learning algorithm, NEWQ, is found to yield the best performance characteristics.

Friedman's Permanent Income Hypothesis as an Example of Diagnostic Reasoning

Many recent developments in artificial intelligence (AI) research are relevant for traditional issues in the philosophy of science. One of the developments in AI research we want to focus on in this article is diagnostic reasoning, which we consider to be of interest for the theory of explanation in general and for an understanding of explanatory arguments in economic science in particular. Usually, explanation is primarily discussed in terms of deductive inferences in classical logic. However, in recent AI research it is observed that a diagnostic explanation is actually quite different from deductive reasoning (see, for example, Reiter, 1987). In diagnostic reasoning the emphasis is on restoring consistency rather than on deduction. Intuitively speaking, the problem diagnostic reasoning is concerned with is the following. Consider a description of a system in which the normal behavior of the system is characterized and an observation that conflicts with this normal behavior. The diagnostic problem is to determine which of the components of the system can, when assumed to be functioning abnormally, account for the conflicting observation. A diagnosis is a set of allegedly malfunctioning components that can be used to restore the consistency of the system description and the observation. In this article, this kind of reasoning is formalized and we show its importance for the theory of explanation. We will show how the diagnosis nondeductively explains the discrepancy between the observed and the correct system behavior. The article also shows the relevance of the subject for real scientific arguments by showing that examples of diagnostic reasoning can be found in Friedman's Theory of the Consumption Function (1957). Moreover, it places the philosophical implications of diagnostic reasoning in the context of Mill's aprioristic methodology.

Cooperative communication: Computerware and humanware

This article emphasizes the combined requirements of computer systems and humanistics. In cooperative computing, negotiations can be used as a basic paradigm by which different roles and their requisites can be identified—the facilitator, the mediator, and the negotiator. The negotiation‐cooperation process has a logical sequence of agreements, definition of terms, objectives, mode of operation, common security measures concerning integrity and liability, handling protocols, etc. The cooperation is based on models of the subject and the partners—i.e., a minimum of three models should be matched. The usual methods of human negotiations supported by metacommunication should have a computer‐realizable substitute. All these subjects are outgrowths of recent research in artificial intelligence (knowledge‐based systems) and cognitive psychology; some experiences are reported in the field. However, the main task is human‐oriented—education of people for this new powerful means of coexistence.

Decision Support Systems (DSS): A Knowledge Oriented Approach

The design of decision supports systems (DSS) is an important reserach area within the field of management information systems (MIS). The design is based on models that consider decision making either as an act which takes place whithin a sociotechnical environment (an organization) or as a cognitive process. The decision supports systems' current architecture and conceptual framework seems to present several weaknesses. Recent research in artificial intelligence shows that decision support systems must increasingly incorporate explicit knowledge in order to meet the objectives set by DSS designers. This paper presents a conceptual framework and a DSS architecture based on the processing of knowledge.

The design of a rule-based CAI tutorial

Rule-based systems are a development associated with recent research in artificial intelligence (AI). These systems express their decision-making criteria as sets of production rules, which are declarative statements relating various system states to program actions. For computer-assisted instruction (CAI) programs, system states are defined in terms of a task analysis and student model, and actions take the form of the different teaching operations that the program can perform. These components are related by a set of means-ends guidance rules that determine what the program will do next for any given state. The paper presents the design of a CAI course employing a rule-based tutorial strategy. This design has not undergone the test of full implementation; the paper presents a conceptual design rather than a programming blueprint. One of the unique features of the course design described here is that it deals with the domain of computer graphics. The precise subject of the course is ReGIS, the Remote Graphics Instruction Set on Digital Equipment Corporation GIGI and VT125 terminals. The paper describes the course components and their inter-relationships, discusses how program control might be expressed in the form of production rules, and presents a program that demonstrates one facet of the intended course: the ability to parse student input in such a way that rules can be used to update a dynamic student model.

Towards self-adaptive interface systems

This paper follows a trend towards more user oriented design approaches to interactive computer systems. The implicit goal in this trend is the development of more “natural” systems. Design approaches should aim at a system capable of continuous change by means of suitable agents. The term “soft facade” is used to describe a modifiable interface to a system. Facades vary in softness and agents for change can be the systems designer, the user or an expert system in the facade. In the latter case, the system is called a self-adaptive interface system. The conditions where a self-adaptive interface system is desirable are briefly considered and discussed in relation to a simple example. Recent research in artificial intelligence is mentioned as a possible source of proposed components for a self-adaptive interface system.

Integrating Knowledge Sources for Computer "Understanding" Tasks

Recent research in artificial intelligence has identified a number of knowledge sources which appear to be needed for effective automatic "understanding" of connected natural language speech and text. These include word-and phrase-level semantics, models for actors and objects, and inference techniques for using script- or goal-oriented knowledge structures. A unified model of the understanding process can be defined using the distributed-computing viewpoint, provided some way can be found to integrate and control a collection of "experts," each one associated with a certain kind of knowledge source. A technique is described, called hierarchical task management, for constructing computer language-processing systems comprising an arbitrary number of distinct, potentially distributed, processes. The technique is based upon the repeated activation and expansion of data structures called tasks, which define important components of the understanding process in terms of a controlled interaction among the experts. The tasks are maintained on several "agendas," and are manipulated by a uniform monitor called the task manager. The process of task management is illustrated in a multiprocess story understander called a distributable script applier mechanism (DSAM), which reads and summarizes newspaper stories about plane crashes.

Data frame model for the engineering design process

Computers are being integrated into the technological design process, just as they are being integrated into banking practice, airline reservations, and many other activities. Until now, computers have served as tools used by designers in methods dictated by the paper-based information systems to which humanity has become accustomed. Computers perform design calculations or analyses, and produce documentation such as drawings, schedules, specifications, and tapes to control production machines. However, as more technical information is stored in computer memories, and as the communication networks between computers develop, different working methods, adapted to electronics-based rather than paper-based information systems, may develop. This paper explores briefly the process of design, and then, using some concepts published in the last few years, proposes a model of the design process that can be implemented by computer. Emphasis is placed on the ‘frame’ concept with a final reference to recent research in artificial intelligence. The paper does not attempt to give a comprehensive survey of the subject of problem solving in engineering design. Furthermore, the ideas expressed have not yet been used to create a working design program. The examples used as illustrations of various concepts are therefore only partial.

Equivalence between AND/OR graphs and context-free grammars

Recent research in artificial intelligence has led to AND/OR graphs as a model of problem decomposition (Nilsson [3]; Simon and Lee [4]). However, AND/OR graphs of a restricted type are equivalent to context-free grammars. This can be set-up formally (the beginnings of a formalism of AND/OR graphs is contained in [4]), but the formalism is so obvious that a brief discussion and example suffice.