Simple Decision-making Model Research Articles

Hierarchical Model-Based Deep Reinforcement Learning for Single-Asset Trading

We present a hierarchical reinforcement learning (RL) architecture that employs various low-level agents to act in the trading environment, i.e., the market. The highest-level agent selects from among a group of specialized agents, and then the selected agent decides when to sell or buy a single asset for a period of time. This period can be variable according to a termination function. We hypothesized that, due to different market regimes, more than one single agent is needed when trying to learn from such heterogeneous data, and instead, multiple agents will perform better, with each one specializing in a subset of the data. We use k-meansclustering to partition the data and train each agent with a different cluster. Partitioning the input data also helps model-based RL (MBRL), where models can be heterogeneous. We also add two simple decision-making models to the set of low-level agents, diversifying the pool of available agents, and thus increasing overall behavioral flexibility. We perform multiple experiments showing the strengths of a hierarchical approach and test various prediction models at both levels. We also use a risk-based reward at the high level, which transforms the overall problem into a risk-return optimization. This type of reward shows a significant reduction in risk while minimally reducing profits. Overall, the hierarchical approach shows significant promise, especially when the pool of low-level agents is highly diverse. The usefulness of such a system is clear, especially for human-devised strategies, which could be incorporated in a sound manner into larger, powerful automatic systems.

437-P: Establishing a Simple Decision-Making Model for the Differential Diagnosis of Diabetes Kidney Disease and Nondiabetes Kidney Disease in T2D Patients through Machine Learning

Objective: To study the 2-DM patients with renal biopsy, analyze the risk factors of DKD and establish a simple decision model to differentiate DKD and NDKD. Methods: 99 2-DM patients were admitted to hospital in Beijing from December 1, 2018 to December 31, 2021. All of the patients accepted kidney biopsy and biochemical test, fundus photography and OCTA. According to the pathological reports, 99 cases were divided into DKD group and NDKD group. The difference indicators of the two groups were compared. The risk factors of DKD were analyzed by logistic regression. A simple and convenient method to distinguish DKD and NDKD was established through the decision tree model. Results: This study was a cross-sectional study. Compared with DKD and NDKD groups, the ratio of male/female in DKD group was higher than that in NDKD group (P=0.033), DR prevalence rate was significantly higher than that in NDKD group (P<0.001), course DM, FBG, HBAIC, SBP, TC and EGFR were also higher than those in NDKD group, and HGB decreased. Binary logistic regression analysis was conducted to determine whether DKD was combined. After adjusting for the above indicators, age, DR, course of DM, HBA1C and HGB were still independently related to DKD. A decision tree model was established with the training set: the course DM, DR, and HBA1C were risk factors of DKD. The accuracy of prediction classification is 75%, the sensitivity was 46.2%, and the specificity was 88.9%. Conclusion: If the patients with type 2 diabetes kidney injury have poor blood glucose control, no DR but anemia, a renal puncture should be considered for definite diagnosis of NDKD. Disclosure L.Zhang: None. Funding Beijing Science & Technology Development of Traditional Chinese Medicine (JJ-2020-25)

Simple decision-making model for orchard air-assisted spraying airflow

Airflow speed is one of the three factors of air-assisted spraying. Optimizing the matching model between airflow speed and target canopy characteristics is an effective way to improve the orchard precision spraying technology, as airflow can significantly affect droplet deposition and drift loss. A simple model of airflow speed was established in this study. First, air-assisted spraying experiments were carried out on a standard simulation canopy to study the airflow speed depended on canopy width, leaf area index, and porosity rate. Second, determined by Ribbon Method and verified by droplet drift data, the airflow speed through the canopy was between 0.5 m/s and 0.7 m/s. Third, multiple tests were carried out under standard simulation canopy with different characteristics, and the airflow speed model was established ultimately: with a fixed leaf area index (LAI), the relationship between canopy upwind boundary airflow speed and canopy width satisfied the exponential model (y=aebx), and the coefficients a and b are well related to the density of branches and leaves in the canopy. When LAI=3.456, y=2.036e1.5887x, R2=0.994; LAI=1.728, y=1.639e1.445x, R2=0.972. Orchard growers can acquire needed airflow speed through this simple model, it is quick and precise and appropriate to most growth periods of a variety of fruit trees, such as apples, pears, and vines. Keywords: airflow speed, canopy width, porosity rate, LAI, air-assisted spraying DOI: 10.25165/j.ijabe.20231602.6849 Citation: Wang X, Feng Y R, Fu W, Qi J T, Song J L. Simple decision-making model for orchard air-assisted spraying airflow. Int J Agric & Biol Eng, 2023; 16(2): 23-29.

Analysis and decision based on specialist self-assessment for prognosis factors of acute leukemia integrating data-driven Bayesian network and fuzzy cognitive map.

The purpose of the present study is to analyze the prognostic factors of acute leukemia and to construct a decision model based on a causal relationship between the factors of this disease to assist medical specialists. In medical decisions, to reach effective, quick, and reliable results, there is a need for a simple decision-making model based on a specialist's self-assessment. It may help the medical team before final diagnosis by costly and time-consuming procedures such as bone marrow sampling and pathological test as well as provide an appropriate prognosis and diagnosis tool. Because of the complex and not the well-defined structure of medical data, the use of intelligent methods must be considered. For this purpose, first, a data-driven Bayesian network (BN) and Greedy algorithm are employed to determine causal relationships and probability between nodes using the real set of data. Then, these causal relationships will form based on the fuzzy cognitive map (FCM). Finally, according to scenarios defined, the results are analyzed. These analyses are also repeated for each type of acute leukemia including acute lymphocytic leukemia (ALL) and acute myelocytic leukemia (AML). Graphical abstract.

Fuzzy Decision Trees in Medical Decision Making Support Systems

Fuzzy decision trees represent classification knowledge more naturally to the way of human thinking and are more robust in tolerating imprecise, conflict, and missing information. Decision Making Support Systems are used widely in clinical medicine because decisions play an important role in diagnostic processes. Decision trees are a very suitable candidate for induction of simple decision-making models with the possibility of automatic learning. The goal of this paper is to demonstrate a new approach for predictive data mining models in clinical medicine. This approach is based on induction of fuzzy decision trees. This approach allows us to build decision-making modesl with different properties (ordered, stability etc.). Three new types of fuzzy decision trees (non-ordered, ordered and stable) are considered in the paper. Induction of these fuzzy decision trees is based on cumulative information estimates. Results of experimental investigation are presented. Predictive data mining is becoming an essential instrument for researchers and clinical practitioners in medicine. Using new approaches based on fuzzy decision trees allows to increase the prediction accuracy. Decision trees are a very suitable candidate for induction using simple decision-making models with the possibility of automatic and AI learning.

Attack and Defense in Cellular Decision-Making: Lessons from Machine Learning

Machine learning algorithms can be fooled by small well-designed adversarial perturbations. This is reminiscent of cellular decision-making where ligands (called antagonists) prevent correct signalling, like in early immune recognition. We draw a formal analogy between neural networks used in machine learning and models of cellular decision-making (adaptive proofreading). We apply attacks from machine learning to simple decision-making models, and show explicitly the correspondence to antagonism by weakly bound ligands. Such antagonism is absent in more nonlinear models, which inspired us to implement a biomimetic defence in neural networks filtering out adversarial perturbations. We then apply a gradient-descent approach from machine learning to different cellular decision-making models, and we reveal the existence of two regimes characterized by the presence or absence of a critical point for the gradient. This critical point causes the strongest antagonists to lie close to the decision boundary. This is validated in the loss landscapes of robust neural networks and cellular decision-making models, and observed experimentally for immune cells. For both regimes, we explain how associated defence mechanisms shape the geometry of the loss landscape, and why different adversarial attacks are effective in different regimes. Our work connects evolved cellular decision-making to machine learning, and motivates the design of a general theory of adversarial perturbations, both for in vivo and in silico systems.

Issues of Confidentiality and Public Interest: A Case Study from a Library Perspective

In 2004, Zimbabwe suffered a number of undocumented challenges, including the externalisation of billions of United States (US) dollars. This was associated with the buying of assets abroad with foreign currency acquired in Zimbabwe. A number of prominent public figures were implicated in these transactions, leading to investigations which did not spare libraries from revealing information about patrons. Libraries connected to institutions, such as the police, hospitals and parliament, have information that is of interest to researchers and, like any other library, also store patron records, which are considered to be of public interest. This article reports on a study that explored a court case where information was provided by the Parliament of Zimbabwe library. The study used a case study methodology and reviewed information in the literature relating to confidentiality and public interest issues, as well as the use of a decision-making model and documents to identify existing operating procedures, if any. The study pointed to the debatable issue of public interest and the need for clear laws on the confidentiality and privacy of patrons in Zimbabwe. It also highlighted the importance for Library and Information Science (LIS) practitioners to understand the legal issues relating to the confidentiality of patron records. Accordingly, clear guidelines are important for decisionmaking when such practitioners are faced with the need to provide patron records. A simple decision-making model is thus recommended to complement existing legislation. Although the story of the court case appeared in local papers in Zimbabwe, placing the case study in a research context provides a useful tool for LIS students and practitioners.

Experiments on partnership and decision making in a disaster environment

The time of entry and the desired activity into a disaster area play an important role in leading to the optimal mix of partners for an agency operating in a disaster scenario. In this paper, based on current literature and interviews in disaster environments by the authors, we designed and executed six experiments. Each experiment is designed to measure a particular component of how agencies make partnership decisions in a disaster environment, with the goal of developing a model of how such decisions would affect operational efficacy. In each experiment, players made partnership and resource allocation decisions in a simulated disaster environment. A wide range of experimental data were gathered to understand how different information influenced decision-making. Each trial of the six experiments had random parameter values, and a logic component for how other agencies in the disaster environment acted with stochastic perturbation. Based on the results of each experiment, we developed a simple model that used the most significant drivers of participant performance. The significant factors in decision making during simulated disaster operations included: agency efficiency, past project investment, partner size, significance of impact on the population, and the amount of remaining need. The impact of this work is two-fold: 1) Identifying some of the underlying biases in human decision making when engaging in cooperative competition, or “coopetition,” and 2) Providing simple decision-making models for understanding, simulating, and predicting agency behavior as part of a disaster response operation.

Bounded Rationality on Consumer Purchase Decisions and Product Diffusion Under Network Externalities: A Study Using Agent-Based Simulation and Experiments with Human Subjects

This study addresses mechanisms of purchase decision-making under network externalities. In particular, we examine how bounded rationality affects product diffusion in markets where network externalities are present. In response to rapid development of information communication technology, markets with network externalities have been expanding recently. If network externalities are present, then a consumer’s utility depends on the number of same product users. In other words, the product value is determined through consumers’ interaction in society. Therefore, the elucidation of the mechanism is required. For this purpose, this study adopts an integral approach with agent-based simulations and experiments with human subjects. We use a simple decision-making model constructed by Ueda et al. [1] to conduct experiments with human subjects based on the method of experimental economics. Additionally, we construct agents who make the bounded rational actions observed in the experiments and examine how product diffusion emerges through interaction of such bounded rational agents. Results show that bounded rationality of subsequent decision-makers affects the purchase behavior of preceding consumers. Accordingly, the output of product diffusion is varied.

Producer decision-making in markets with network externalities

Markets with network externalities, by which more users make a product more valuable, such as mobile telephone and Internet service markets, have been growing rapidly. In such markets, the value of products is created through dynamic interaction among consumers and producers. This paper presents analyses of the mechanisms of their interaction in a market with network externalities, particularly addressing producers’ decision-making. A simple decision-making model is constructed and examined using multi-agent simulation, game theoretic analysis, and experiments using human subjects. Results show that incomplete information among consumers might encourage new product market growth. Additionally, we discuss producer strategies for a product introduction plan, comparing results with real world data.

Supervision du réseau d’assainissement de Marseille et gestion des épisodes pluvieux

Click to increase image sizeClick to decrease image sizeFaced to the rainfall risks, the City of Marseilles proceeds with the forecasting and the real-time management of the sewer network and of field action plans. The main purpose is to anticipate and to activate before the rain starts an operational plan of information and of ground actions and then to manage it during the rain. Depending on the rain severity, the management gives priority to different fields : In the event of light rainfall – in order to preserve the quality of sea-water for bathing, thus avoiding sewer discharges.In the event of average rainfall – in order to avoid surface flooding in the city, thus giving priority to sewer overflows towards the sea.In the event of heavy rainfall – in order to ensure public safety, thus managing assistance plan by the municipality services.Therefore, relationship between rainfall and severity levels were established for the city, considering the data on rainfall and its potential effects on the sewerage system, flooding points and also the number of field interventions due to phone calls. Finally, risk management is based on a simple decision-making model, using as input, predicted data, measured rainfall and real water level measurements. Non-dangerous rainfalls are managed by the sewerage system exploitant and dangerous rainfalls by local authorities.

Incorporating ecological sustainability into landscape planning

The ecological component is crucial in landscape planning according to the principles of sustainable development. We define “ecologically sustainable landscape” and develop a tool to measure how ecological sustainability is incorporated in landscape plans. This method acknowledges the critical role of spatial scale and pattern to the conservation of biodiversity. The metapopulation concept is used as a spatially explicit ecological theory, appropriate to describe the relation between biodiversity and the pattern of ecosystem patches (“ecosystem network”) in intensively used regions. We propose that ecological sustainability is achieved if quality, area and configuration of the ecosystem network permit target species to persist. A simple decision-making model represents a theoretical framework for a tool comprising two sets of ecological indicators. One set indicates the awareness of actors to consider ecological principles of sustainable planning. The other set indicates their performance to apply these principles quantitatively in designing the ecosystem pattern. The method is applied on a sample of reports on Dutch landscape development plans. A majority of the reports shows awareness of the importance of spatial conditions for achieving planning goals, but perform inadequately on the quantitative indicators. We conclude that the tool could be developed as a guideline and assessment method for the ecological sustainability of landscape plans.

Judging Transitional Justice

Truth revelation procedures are evaluated according to various normative criteria. The authors find the concepts of false conviction and false acquittal more adequate for such evaluation than the conformity with the rule of law and apply a useful classification of truth revelation procedures into incentive-based (ITRs) and evidence-based ones (ETRs). ITRs induce perpetrators and secret agents of the authoritarian regime to reveal the truth about their past, while ETRs rely exclusively on preserved evidence and victims’ testimonies. Using a simple decision-making model, the authors show that while both procedures are sensitive to the problem of falsified evidence, ITRs perform better with respect to revealing the identity of collaborators whose files were destroyed. Finally, they discuss the connection between ITRs and two modes of coming to terms with the past, endogenous and exogenous.

The Economic Value Of Ensemble-Based Weather Forecasts

Abstract The potential economic benefit associated with the use of an ensemble of forecasts versus anequivalent or higher-resolution control forecast is discussed. Neither forecast systems are post-processed,except a simple calibration that is applied to make them reliable. A simple decision-making model is used where all potential users of weather forecasts are characterized by the ratio between the cost of their action to preventweather-related damages, and the loss that they incur in case they do not protect their operations. It isshown that the ensemble forecast system can be used by a much wider range of users. Furthermore,for many, and for beyond 4-day lead time for all users, the ensemble provides greater potential economicbenefit than a control forecast, even if the latter is run at higher horizontal resolution. It is argued that theadded benefits derive from 1) the fact that the ensemble provides a more detailed forecast probabilitydistribution, allowing the users to tailor their weather forecast...

Supervisory Control of a Flexible Manufacturing System

This paper describes a study that explored human capabilities in supervisory control of a flexible manufacturing system (FMS). Implications of current design strategies concerning the human supervisory role in FMSs are discussed. Also examined are issues in human supervisory control that arise from features characteristic of these types of automated systems. A real-time interactive simulation methodology is described that served as the basis for evaluating human performance. Results are explained in terms of two frameworks: (1) one that incorporated hypothetical dimensions of the mental workload construct into various task conditions; and (2) a simple decision-making model intrinsic to which was the imposition on the human of a relatively well-defined locus of control. The effectiveness of graphics was evaluated as a decision-support device aimed at enhancing human pattern-recognition capabilities.