Design Of Agents Research Articles

Co-Designing with Users the Explanations for a Proactive Auto-Response Messaging Agent

Explanations of AI Agents' actions are considered to be an important factor in improving users' trust in the decisions made by autonomous AI systems. However, as these autonomous systems evolve from reactive, i.e., acting on user input, to proactive, i.e., acting without requiring user intervention, there is a need to explore how the explanation for the actions of these agents should evolve. In this work, we explore the design of explanations through participatory design methods for a proactive auto-response messaging agent that can reduce perceived obligations and social pressure to respond quickly to incoming messages by providing unavailability-related context. We recruited 14 participants who worked in pairs during collaborative design sessions where they reasoned about the agent's design and actions. We qualitatively analyzed the data collected through these sessions and found that participants' reasoning about agent actions led them to speculate heavily on its design. These speculations significantly influenced participants' desire for explanations and the controls they sought to inform the agents' behavior. Our findings indicate a need to transform users' speculations into accurate mental models of agent design. Further, since the agent acts as a mediator in human-human communication, it is also necessary to account for social norms in its explanation design. Finally, user expertise in understanding their habits and behaviors allows the agent to learn from the user their preferences when justifying its actions.

4-Hydroxybenzoic Acid and β-Sitosterol from Cucurbita ficifolia Act as Insulin Secretagogues, Peroxisome Proliferator-Activated Receptor-Gamma Agonists, and Liver Glycogen Storage Promoters: In Vivo, In Vitro, and In Silico Studies.

Insulin secretion and GLUT4 expression are two critical events in glucose regulation. The receptors G-protein-coupled receptor 40 (GPR40) and peroxisome proliferator-activated receptor-gamma (PPARγ) modulate these processes, and they represent potential therapeutic targets for new antidiabetic agent's design. Cucurbita ficifolia fruit is used in traditional medicine for diabetes control. Previous studies demonstrated several effects: a hypoglycemic effect mediated by an insulin secretagogue action, antihyperglycemic effect, and promoting liver glycogen storage. Anti-inflammatory and antioxidant effects were also reported. Moreover, some of its phytochemicals have been described, including d-chiro-inositol. However, to understand these effects integrally, other active principles should be investigated. The aim was to perform a chemical fractionation guided by bioassay to isolate and identify other compounds from C. ficifolia fruit that explain its hypoglycemic action as insulin secretagogue, its antihyperglycemic effect by PPARγ activation, and on liver glycogen storage. Three different preparations of C. ficifolia were tested in vivo. Ethyl acetate fraction derived from aqueous extract showed antihyperglycemic effect in an oral glucose tolerance test and was further fractioned. The insulin secretagogue action was tested in RINm5F cells. For the PPARγ activation, C2C12 myocytes were treated with the fractions, and GLUT4 mRNA expression was measured. Chemical fractionation resulted in the isolation and identification of β-sitosterol and 4-hydroxybenzoic acid (4-HBA), which increased insulin secretion, GLUT4, PPARγ, and adiponectin mRNA expression, in addition to an increase in glycogen storage. 4-HBA exhibited an antihyperglycemic effect, while β-sitosterol showed hypoglycemic effect, confirming the wide antidiabetic related results we found in our in vitro models. An in silico study revealed that 4-HBA and β-sitosterol have potential as dual agonists on PPARγ and GPR40 receptors. Both compounds should be considered in the development of new antidiabetic drug development.

Evaluating Gin Rummy Hands Using Opponent Modeling and Myopic Meld Distance

Gin Rummy is a popular two-player card game involving choices to draw and discard cards to form sets of matching cards. Unlike other popular games such as Chess, Poker, and Go, there is little formal artificial intelligence research about how to make good decisions when playing Gin Rummy. In this paper, we develop an agent that plays Gin Rummy through a combination of known and expected card values, modeling the opponent to predict their cards of interest, and a conservative approach to assessing when to end the hand. In addition to discussing our observations about Gin Rummy that inspired our agent's design and how the agent works, we evaluate the relative importance of various features employed by our agent by competing agents which implement various subsets of those features.

The Effect of Valence and Arousal on Virtual Agent's Designs in Quiz Based Multimedia Learning Environment

The Effect of Valence and Arousal on Virtual Agent's Designs in Quiz Based Multimedia Learning Environment

Bubble Inflation Using Phase-Change Perfluorocarbon Nanodroplets as a Strategy for Enhanced Ultrasound Imaging and Therapy.

Phase-change perfluorocarbon microdroplets were introduced over 2 decades ago to occlude downstream vessels in vivo. Interest in perfluorocarbon nanodroplets has recently increased to enable extravascular targeting, to rescue the weak ultrasound signal of perfluorocarbon droplets by converting them to microbubbles and to improve ultrasound-based therapy. Despite great scientific interest and advances, applications of phase-change perfluorocarbon agents have not reached clinical testing because of efficacy and safety concerns, some of which remain unexplained. Here, we report that the coexistence of perfluorocarbon droplets and microbubbles in blood, which is inevitable when droplets spontaneously or intentionally vaporize to form microbubbles, is a major contributor to the observed side effects. We develop the theory to explain why the coexistence of droplets and microbubbles results in microbubble inflation induced by perfluorocarbon transfer from droplets to adjacent microbubbles. We also present the experimental data showing up to 6 orders of magnitude microbubble volume expansion, which occludes a 200 μm tubing in the presence of perfluorocarbon nanodroplets. More importantly, we demonstrate that the rate of microbubble inflation and ultimate size can be controlled by manipulating formulation parameters to tailor the agent's design for the potential theranostic application while minimizing the risk to benefit ratio.

Reinforcement Learning for Improving Agent Design.

In many reinforcement learning tasks, the goal is to learn a policy to manipulate an agent, whose design is fixed, to maximize some notion of cumulative reward. The design of the agent's physical structure is rarely optimized for the task at hand. In this work, we explore the possibility of learning a version of the agent's design that is better suited for its task, jointly with the policy. We propose an alteration to the popular OpenAI Gym framework, where we parameterize parts of an environment, and allow an agent to jointly learn to modify these environment parameters along with its policy. We demonstrate that an agent can learn a better structure of its body that is not only better suited for the task, but also facilitates policy learning. Joint learning of policy and structure may even uncover design principles that are useful for assisted-design applications.

A multi-agent system for optimal sizing of a cooperative self-sustainable multi-carrier microgrid

In this paper, an interactive multi-agent system (MAS) is applied to the problem of optimal sizing of a cooperative self-sustainable multi-carrier microgrid that includes various privately-owned entities. The proposed microgrid includes photovoltaic (PV) arrays, batteries, an electrolyzer, a hydrogen tank, a fuel cell (FC), a reactor-reformer system, a hydrogen compressor-dispenser system, a converter, residential electrical loads, and a charging/refilling station. The proposed MAS enables information exchange required for the application of demand-side management (DSM) strategy and has five agents, namely generation agent (GA), electrical load agent (LA), charging/refilling station agent (SA), control agent (CA), and design agent (DA). The GA is responsible for managing the distributed energy resources of the microgrid. The LA aggregates the residential electrical loads. The SA is responsible for charging of plug-in hybrid electric vehicles (PHEVs) and refilling of fuel cell electric vehicles (FCEVs). The CA coordinates the interactions between the field level agents. The DA finds the optimal sizes of the system’s components by minimizing the total cost of the system through particle swarm optimization (PSO) algorithm. Simulation results demonstrate that the proposed system can reduce the overall cost of the microgrid in comparison with non-interactive methods.

Planning-Based Social Partners for Children with Autism

This paper describes the design and implementation of a planning-based socially intelligent agent built to help young children with Autism Spectrum Conditions acquire social communication skills. We explain how planning technology allowed us to satisfy the requirements relating to the agent's design that we identified through our consultations with children and carers as well as our review of best practices for autism intervention. We discuss the design principles we implemented, the engineering challenges we faced and the lessons we learned in building the pedagogical agent. We conclude by presenting extensive experimental results concerning the agent's efficacy.

Strategic Information Disclosure to People with Multiple Alternatives

This paper studies how automated agents can persuade humans to behave in certain ways. The motivation behind such agent's behavior resides in the utility function that the agent's designer wants to maximize and which may be different from the user's utility function. Specifically, in the strategic settings studied, the agent provides correct yet partial information about a state of the world that is unknown to the user but relevant to his decision. Persuasion games were designed to study interactions between automated players where one player sends state information to the other to persuade it to behave in a certain way. We show that this game theory based model is not sufficient to model human-agent interactions, since people tend to deviate from the rational choice. We use machine learning to model such deviation in people from this game theory based model. The agent generates a probabilistic description of the world state that maximizes its benefit and presents it to the users. The proposed model was evaluated in an extensive empirical study involving road selection tasks that differ in length, costs and congestion. Results showed that people's behavior indeed deviated significantly from the behavior predicted by the game theory based model. Moreover, the agent developed in our model performed better than an agent that followed the behavior dictated by the game-theoretical models.

A Price-Based Negotiation Mechanism for Distributed Collaborative Design

With an increased focus on collaborative design for success in global markets, there is a need for a negotiation mechanism for designs distributed over autonomous stakeholders. This paper utilizes the unit price of the design variables as a medium for negotiation between autonomous and self-interested design participants represented by design agents (DAs). An artificial agent - the facilitator agent (FA), maximizes the overall utility of the design as a function of the prices of the design variables, and guides the design toward global feasibility. By relating the FA's price space to the DAs' design space through price-variable curves, a negotiation platform is developed and tested.

Preferencias individuales y elección social. La inocencia epistémica de las teorías

The economic approach, that actually prevail in the social choice studies, have no want consequence when is used in choice of possible social states. The basic reason is that economic model proscribe any axiological consideration in this field, removing in agent's design, ethics and political consideration and any reference about individual's values in the take of collective decisions process.

A contradiction-driven approach to theory formation: conceptual issues, pragmatics in human learning, potentialities

In Educational literature, Discovery Learning appears as an approach in which the learner builds up his/her own knowledge by performing experiments within a domain and inferring/increasing rules as a result. Such a constructivist approach has been largely exploited in the design of computational artefacts with learning purposes, the so-called Discovery Learning Environments (DLEs). One known feature of such environments is the autonomy degree required for students to succeed while handling a domain. Additionally, DLEs designers are often challenged to get students actually engaged. Such questions are on the basis of our concerns with the design and usage of particular DLEs, within which learning events occur as a consequence of contradiction detection and overcoming, during human/machine cooperative work. In this paper, we present an artificial agent capable of handling such a contradiction-driven approach of learning, by highlighting the exchanges that the agent should promote with a human learner. The conceptual model supporting the agent's design relies on the scientific rationale, particularly the empirical approach guided by the theory-experiment confrontation. We shall reinforce the interest of the model for the design of DLEs by presenting its exploitation in a real learning situation in Law. Also, we suggest potential instantiations of the model elsewhere than in Human Learning.