AI-powered educational technologies have demonstrated measurable benefits for learners, but their design and evaluation have largely centered on K-12 contexts. As a result, many AI-supported learning systems remain poorly aligned with the needs, constraints, and goals of adult learners. To better understand how AI systems function in adult education, this paper examines the deployment of several AI learning technologies developed within a multidisciplinary, national research institute in the United States focused on adult learning and online education. Drawing on longitudinal deployment data, we conducted a reflexive thematic analysis to identify recurring challenges and design considerations across systems. These insights were synthesized into a set of 19 design guidelines intended to inform future AI-supported adult learning technologies. We demonstrate the utility of these guidelines through a heuristic evaluation of the deployed systems. Lastly, we present a guideline exploration tool that aids in the ideation of technologies by connecting the guidelines to stakeholder statements surfaced in the analysis process.

The ability to correct mistakes and adapt to users' changing needs is critical for AI agents to remain robust and trustworthy. LLM-based agents are inherently prone to errors like hallucinations and misinterpretations. We observed this challenge in SAMI, an AI social agent deployed in Georgia Tech's OMSCS program for ten semesters (11,000+ users). Users frequently requested the agent to revise its knowledge base, both to correct LLM-induced errors and to update their information. To support such revisions, we introduce a two-level metacognitive self-adaptation architecture that integrates knowledge-based AI (KBAI) with LLMs. The architecture comprises a cognitive layer that performs the agent's core tasks, and a metacognitive layer that introspects on the cognitive layer's process using a Task–Method–Knowledge (TMK) model of the agent. The metacognitive layer identifies the task that needs revision, updates the knowledge base, and communicates the revision process to the user.

Recent advances in AI are integrating AI into the fabric of human social life, creating transformative, co-shaping relationships between humans and AI. This trend makes it urgent to investigate how these systems, in turn, shape their users. We conducted a three-phase design study with 24 participants to explore this dynamic. Our findings reveal critical tensions: (1) social AI often exacerbates the very interpersonal problems it is designed to mitigate; (2) it introduces nuanced privacy harms for secondary users inadvertently involved in AI-mediated social interactions; and (3) it can threaten the primary user’s personal agency and identity. We argue these tensions expose a problematic tendency in the user-centered paradigm, which often prioritizes immediate user experience at the expense of core human values like interpersonal ethics and self-efficacy. We call for a paradigm shift toward a more provocative and relational design perspective that foregrounds long-term social and personal consequences.

Scalable AI tutoring for procedural skill learning requires structured knowledge representations, yet constructing these representations remains a labor-intensive bottleneck. This paper introduces a new LLM-assisted text-to-model (TTM) methodology that transforms instructional materials into schema-complete Task-Method-Knowledge (TMK) models through ontology-constrained prompting and template-based generation, automating structural scaffolding while preserving expert oversight. Applied to a graduate-level online AI course, the methodology produced 23 TMK models---enabling full-course coverage for Ivy, a deployed AI coach that relies on TMK models to support learners' procedural understanding, for the first time. AI-assisted authoring reduced expert modeling time by 50-70% while producing structurally valid and highly reproducible models. We evaluate structural validity, semantic alignment, reproducibility, and refinement effort to characterize authoring scalability. Results indicate that the TTM methodology substantially lowers the cost of constructing structured procedural representations, making course-wide deployment of structured AI tutoring systems practically feasible.

Centering humans in explainable artificial intelligence (XAI) research has primarily focused on AI model development and highstake scenarios.However, as AI becomes increasingly integrated into everyday applications in often opaque ways, the need for explainability tailored to end-users has grown more urgent.To address this gap, we explore end-users' perspectives on embedding XAI into daily AI application scenarios.Our findings reveal that XAI is not naturally accepted by end-users in their daily lives.When users seek explanations, they envision XAI design that promotes contextualized understanding, empowers adoption and adaption to AI systems, and considers multistakeholders' values.We further discuss supporting users' agency in XAI non-use and alternatives to XAI for managing ambiguity in AI interactions.Additionally, we provide design implications for XAI design at personal and societal levels.These include understanding users through a computational rationality lens, adaptive design that coevolves with users, and advancing the "society-in-the-loop" vision with everyday XAI.

Modular AI systems can be developed using LLM-prompts-based modules to minimize deployment time even for complex tasks.However, these systems do not always perform well and improving them using the data traces collected from a deployment remains an open challenge.The data traces contain LLM inputs and outputs, but the annotations from LLMs are noisy.We hypothesize that Active Label Correction (ALC) can be use on the collected data to train smaller task-specific improved models that can replace LLM-based modules.In this paper, we study the noise in three GPT-3.5annotateddatasets and their denoising with human feedback.We also propose a novel method ALC3 that iteratively applies three updates to the training dataset: auto-correction, correction using human feedback and filtering.Our results show that ALC3 can lead to oracle performance with feedback on 17-24% fewer examples than the number of noisy examples in the dataset across three different NLP tasks.

Online learning at scale has become dramatically more popular over the last decade. While these programs provide affordable and accessible education, low retention and engagement are persistent problems. Virtual Teaching Assistants (VTAs) offer a solution: VTAs such as Jill Watson can answer questions about course logistics and content, amplifying interaction between professors and students, increasing teaching presence, and thereby improving retention and engagement. Using the Community of Inquiry framework, this paper presents what we believe is the first experimental study of the effect a VTA has on student perceptions of teaching presence, social presence, and cognitive presence. Students in a large, online, graduate computer science course were randomly assigned to sections with and without access to Jill Watson. The Community of Inquiry survey was then administered at the end of the semester to measure the three presences. We find that Jill Watson has a small, positive, statistically significant effect on the Design & Organization dimension of teaching presence as well as social presence.

A/B experiments are commonly used in research to compare the effects of changing one or more variables in two different experimental groups-a control group and a treatment group. While the benefits of using A/B experiments are widely known and accepted in education, there is less agreement on an approach to creating software infrastructure systems to assist in rapidly conducting such experiments in the field. To assist in alleviating this gap, we are creating a software infrastructure for A/B experiments that allows researchers to conduct experiments and automatically analyze their results for an education-focused ecology-based conceptual modeling platform.

Due to the lack of face-to-face interactions, online learners frequently experience social isolation that negatively impacts students’ well-being and learning experiences. Many text-based AI agents have been equipped with different social characteristics and functionalities to support people who are socially isolated. However, the design of agent’s functionalities, social characteristics, and ethical challenges in promoting social connectedness among online learners are underexplored. Taking a co-design approach, we included 23 online learners enrolled in an online for-degree graduate program as active participants in two virtual co-design workshop studies. Through four different co-design activities, we identified online learners’ preferences for AI agent’s functionalities and social characteristics in promoting their social connectedness as well as potential ethical concerns. Based on our findings, we establish the role of AI agent as a facilitator to continuously scaffold online learners’ social connection process. We further discuss the unique ethical challenges regarding agent-mediated social interaction in online learning.

When a robot adapts a learned task for a novel environment, any changes to objects in the novel environment have an unknown effect on its task execution. For example, replacing an object in a pick-and-place task affects where the robot should target its actions, but does not necessarily affect the underlying action model. In contrast, replacing a tool that the robot will use to complete a task will effectively alter its end-effector pose with respect to the robot's base coordinate system, and thus the robot's motion must be replanned accordingly. These examples highlight the relationship among (i) differences between the source and target environments, (ii) the level of abstraction at which a robot's task model should be represented to enable transfer to the target environment, and (iii) the information needed to ground the abstracted task representation in the target environment. In this abstract, summarizing our full article [Fitzgerald et al., 2021], we present our taxonomy of transfer problems based on this relationship. We also describe a knowledge representation called the Tiered Task Abstraction (TTA) and demonstrate its applicability to a variety of transfer problems in the taxonomy. Our experimental results indicate a trade-off between the generality and data requirements of a task representation, and reinforce the need for multiple transfer methods that operate at different levels of abstraction.

Our online interactions are constantly mediated through Artificial Intelligence (AI), especially our social interactions. AI-mediated social interaction is the AI-facilitated process of building and maintaining social connections between individuals through information inferred from people's online posts. With its impending application across a number of contexts, the challenges and opportunities of AI-mediated social interaction remain underexplored. This paper seeks to understand the design space of AI-mediated social interaction in the context of online learning, where students frequently face social isolation. We deployed an AI agent named SAMI in three class discussion forums to help online learners build social connections. Using SAMI as a probe, we conducted semi-structured interviews with 26 students to understand their difficulties in remote social interactions and their experiences with SAMI. Through the lenses of social translucence and social-technical gap, we illustrate online learners' difficulties in remote social interactions and how SAMI resolved some of the difficulties. We also identify potential ethical and social challenges of SAMI such as user agency and privacy. Based on our findings, we outline the design space of AI-mediated social interaction. We discuss the design tension between AI performance and ethical design and pinpoint two design opportunities for AI-mediated social interaction in designing towards human-AI collaborative social matching and artificial serendipity.

Building conversational agents that can conduct natural and prolonged conversations has been a major technical and design challenge, especially for community-facing conversational agents. We posit Mutual Theory of Mind as a theoretical framework to design for natural long-term human-AI interactions. From this perspective, we explore a community’s perception of a question-answering conversational agent through self-reported surveys and computational linguistic approach in the context of online education. We first examine long-term temporal changes in students’ perception of Jill Watson (JW), a virtual teaching assistant deployed in an online class discussion forum. We then explore the feasibility of inferring students’ perceptions of JW through linguistic features extracted from student-JW dialogues. We find that students’ perception of JW’s anthropomorphism and intelligence changed significantly over time. Regression analyses reveal that linguistic verbosity, readability, sentiment, diversity, and adaptability reflect student perception of JW. We discuss implications for building adaptive community-facing conversational agents as long-term companions and designing towards Mutual Theory of Mind in human-AI interaction.

As online education proliferates, one concern that has been raised is that it may fail to capture desirable emergent phe-nomena from on-campus programs. Student community is one example of such a phenomenon: on-campus student communities thrive based on synchronous collocation. An online program might be designed to capture all deliberate constructs in an on-campus program, but there may be beneficial side effects of synchronous collocation that are not apparent. In this work, we examine the issue of social isolation in an online graduate program. By happenstance, three studies were conducted in relative isolation looking at social isolation from different angles. The first study exam-ined trajectories in social presence as a semester proceeded. The second study developed an understanding of students' needs with regard to community in an online program. The third study tested out an immersive virtual environment to try to improve students' sense of connectedness. Combin-ing their findings, we find compelling evidence of the exist-ence of a Synchronicity Paradox in online education: stu-dents desire synchronicity to form strong social communi-ties, and yet part of the chief appeal of these online pro-grams is their asynchronicity. In light of this finding, we provide design guidelines for how synchronicity may be reintroduced into asynchronous programs without sacrific-ing the benefits of asynchronicity. More specifically, we propose that scale itself may be the key to building emer-gent synchronicity.

Graphical models offer techniques for capturing the structure of many problems in real-world domains and provide means for representation, interpretation, and inference. The modeling framework provides tools for discovering rules for solving problems by exploring structural relationships. We present the Structural Affinity method that uses graphical models for first learning and subsequently recognizing the pattern for solving problems on the Raven's Progressive Matrices Test of general human intelligence. Recently there has been considerable work on computational models of addressing the Raven's test using various representations ranging from fractals to symbolic structures. In contrast, our method uses Markov Random Fields parameterized by affinity factors to discover the structure in the geometric analogy problems and induce the rules of Carpenter et al.'s cognitive model of problem-solving on the Raven's Progressive Matrices Test. We provide a computational account that first learns the structure of a Raven's problem and then predicts the solution by computing the probability of the correct answer by recognizing patterns corresponding to Carpenter et al.'s rules. We demonstrate that the performance of our model on the Standard Raven Progressive Matrices is comparable with existing state of the art models.

Jill Watson is our name for a virtual teaching assistant for a Georgia Tech course on artificial intelligence: Jill answers routine, frequently asked questions on the class discussion forum. In this paper, we outline some of the ethical issues that arose in the development and deployment of the virtual teaching assistant. We posit that experiments such as Jill Watson are critical for deeply understanding AI ethics.

In four years, the Georgia Tech Online MS in CS (OMSCS) program has grown from 200 students to over 6000. Despite early evidence of success, there is a need to evaluate the program's effectiveness. In this paper, we focus on trends from Fall 2014 to Fall 2017 in the on-campus and online sections of one OMSCS course, Knowledge-Based Artificial-Intelligence (KBAI). We leverage sentiment analysis and readability assessments to quantify the evolving quality of discourse on the online forum discussions of the various sections. The research was conducted as a longitudinal study, and aims to evaluate the success of the KBAI course by comparing trends between residential and online sections. Despite slight downward trends in online discourse quality and sentiment polarity, our results suggest that the growing OMSCS program has been successful in replicating the quality of learning experienced by on-campus students in the KBAI course.

When transferring a learned task to an environment containing new objects, a core problem is identifying the mapping between objects in the old and new environments. This object mapping is dependent on the task being performed and the roles objects play in that task. Prior work assumes (i) the robot has access to multiple new demonstrations of the task or (ii) the primary features for object mapping have been specified. We introduce an approach that is not constrained by either assumption but rather uses structured interaction with a human teacher to infer an object mapping for task transfer. We describe three experiments: an extensive evaluation of assisted object mapping in simulation, an interactive evaluation incorporating demonstration and assistance data from a user study involving 10 participants, and an offline evaluation of the robot’s confidence during object mapping. Our results indicate that human-guided object mapping provided a balance between mapping performance and autonomy, resulting in (i) up to 2.25× as many correct object mappings as mapping without human interaction, and (ii) more efficient transfer than requiring the human teacher to re-demonstrate the task in the new environment, correctly inferring the object mapping across 93.3% of the tasks and requiring at most one interactive assist in the typical case.

Case-based reasoning addresses new problems by remembering and adapting solutions previously used to solve similar problems. Pulled by the increasing number of applications and pushed by a growing interest in memory intensive techniques, research on case-based reasoning appears to be gaining momentum. In this article, we briefly summarize recent developments in research on case-based reasoning based partly on the recent Twenty Fourth International Conference on Case-Based Reasoning.

In Fall 2014 we offered an online course on Knowledge-Based Artificial Intelligence (KBAI) to about 200 students as part of the Georgia Tech Online MS in CS program. By now we have offered the course to more than 1000 students. We describe the design, development and delivery of the online KBAI class in Fall 2014.

The field of AI has changed significantly in the past couple of years and will likely continue to do so. Driven by a desire to expose our students to relevant and modern materials, we conducted two surveys, one of AI instructors and one of AI practitioners. The surveys were aimed at gathering infor-mation about the current state of the art of introducing AI as well as gathering input from practitioners in the field on techniques used in practice. In this paper, we present and briefly discuss the responses to those two surveys.

Large classes, both online and residential, typically demand many graders for evaluating students' written work. Some classes attempt to use autograding or peer grading, but these both present challenges to assigning grades at for-credit institutions, such as the difficulty of autograding to evaluate free-response answers and the lack of expert oversight in peer grading. In a large, online class at Georgia Tech in Summer 2015, we experimented with a new approach to grading: framing graders as meta-reviewers, charged with evaluating the original work in the context of peer reviews. To evaluate this approach, we conducted a pair of controlled experiments and a handful of qualitative analyses. We found that having access to peer reviews improves the perceived quality of feedback provided by graders without decreasing the graders' efficiency and with only a small influence on the grades assigned.

Many ongoing efforts in online education aim to increase accessibility through affordability and flexibility, but some critics have noted that pedagogy often suffers during these efforts. In contrast, in the low-cost for-credit Georgia Tech Online Masters of Science in Computer Science (OMSCS) program, we have observed that the features that make the program accessible also lead to pedagogical benefits. In this paper, we discuss the pedagogical benefits, and draw a causal link between those benefits and the factors that increase the program's accessibility.

Despite the ubiquitous use of videos in online learning and enormous literature on designing online learning, there has been relatively little research on what pedagogical strategies should be used to make the most of video lessons and what constitutes an effective video for student learning. We experimented with a model of incorporating four pedagogical strategies, four instructional phases, and four production guidelines-in designing and developing video lessons for an online graduate course. In this paper, we share our experience as well as students' perceptions of their effectiveness. We also discuss what needs to be done for future research.

This paper presents the design and evaluation of a set of intelligent tutoring agents constructed to teach teams of students an authentic process of inquiry-driven modeling. The paper first presents the theoretical grounding for inquiry-driven modeling as both a teaching strategy and a learning goal, and then presents the need for guided instruction to improve learning of this skill. However, guided instruction is difficulty to provide in a one-to-many classroom environment, and thus, this paper makes the case that interaction with a metacognitive tutoring system can help students acquire the skill. The paper then describes the design of an exploratory learning environment, the Modeling and Inquiry Learning Application (MILA), and an accompanying set of metacognitive tutors (MILA--T). These tools were used in a controlled experiment with 84 teams (237 total students) in which some teams received and interacted with the tutoring system while other teams did not. The effect of this experiment on teams' demonstration of inquiry-driven modeling are presented.

We report a novel approach to addressing the Raven’s Progressive Matrices (RPM) tests, one based upon purely visual representations. Our technique introduces the calculation of confidence in an answer and the automatic adjustment of level of resolution if that confidence is insufficient. We first describe the nature of the visual analogies found on the RPM. We then exhibit our algorithm and work through a detailed example. Finally, we present the performance of our algorithm on the four major variants of the RPM tests, illustrating the impact of confidence. This is the first such account of any computational model against the entirety of the Raven’s.

Scientists use both conceptual models and executable simulations to help them make sense of the world. Models and simulations each have unique affordances and limitations, and it is useful to leverage their affordances to mitigate their respective limitations. One way to do this is by generating the simulations based on the conceptual models, preserving the capacity for rapid revision and knowledge sharing allowed by the conceptual models while extending them to provide the repeated testing and feedback of the simulations. In this paper, we present an interactive system called MILAfiS for generating agent-based simulations from conceptual models of ecological systems. Designed with STEM education in mind, this user-centered interface design allows the user to construct a Component-Mechanism-Phenomenon conceptual model of a complex system, and then compile the conceptual model into an executable NetLogo simulation. In this paper, we present the results of a pilot study with this interface with about 50 middle school students in the context of learning about ecosystems.

We present two visual algorithms, called the affine and fractal methods, which each solve a considerable portion of the Raven’s Progressive Matrices (RPM) test. The RPM is considered to be one of the premier psychometric measures of general intelligence. Current computational accounts of the RPM assume that visual test inputs are translated into propositional representations before further reasoning takes place. We propose that visual strategies can also solve RPM problems, in line with behavioral evidence showing that humans do use visual strategies to some extent on the RPM. Our two visual methods currently solve RPM problems at the level of typical 9- to 10-year-olds.

Biologically inspired design perhaps is one of the most important movements in engineering design. The paradigm espouses use of analogies to biology in generating conceptual designs for new technologies. In this paper, we briefly summarize some empirical findings about biologically inspired design, and then develop an information-processing theory of creative analogies in biologically inspired design. We also compare our theory with similar theories. In addition, we examine how biologically inspired design is fundamentally different from other design paradigms.

The Odd One Out test of intelligence consists of 3x3 matrix reasoning problems organized in 20 levels of difficulty. Addressing problems on this test appears to require integration of multiple cognitive abilities usually associated with creativity, including visual encoding, similarity assessment, pattern detection, and analogical transfer. We describe a novel fractal technique for addressing visual analogy problems on the Odd One Out test. In our technique, the relationship between images is encoded fractally, capturing inherent self-similarity. The technique starts at a high level of resolution, but, if that is not sufficient to resolve ambiguity, it automatically adjusts itself to the right level of resolution for addressing a given problem. Similarly, the technique automatically starts with searching for similarity between simpler relationships, but, if that is not sufficient to resolve ambiguity, it automatically searches for similarity between higher-order relationships. We present preliminary results from applying the fractal technique on a representative subset of the problems from the Odd One Out test.

An important issue in teaching interdisciplinary biologically inspired design is the external representations we use to foster understanding of biological systems. In this study we explore if functional models of biological systems, and in particular Structure-Behavior-Function (SBF) models, enable humans to better understand complex biological systems. The study compares the use of SBF models in answering questions about biological systems versus the use of textual, tabular and graphical representations. The results indicate that while no one representation is best for answering all types of questions, SBF models enable more accurate answers to questions entailing abstract and complex inferences.

In this paper, we describe the evolution of an interactive technology called the Ecological Modeling Toolkit (EMT) that supports learning about complex ecological systems in middle school science. Authentic learning of science is facilitated by imitation, rehearsal and understanding of real-world scientific practices such as observation, experimentation, problem formulation, hypothesis testing, and model construction and revision. We illustrate how the tools in EMT work together to support many real-world scientific practices such as model construction, simulation and revision, and scaffold others such as observation, problem formulation and hypothesis testing.

The task of biologically inspired design requires designers to access, understand, and apply knowledge of biological systems. One common source for obtaining this kind of knowledge is scholarly biology articles accessed from online libraries and bibliographic databases (e.g., Web of Science, Google Scholar). However, our studies show that such online information environments do not adequately support this kind of interdisciplinary research. Designers need more help with both accessing relevant biology articles and understanding the biological systems that are described in those articles. In this paper, we present Biologue, a social citation cataloging system that uses model-based tagging to address these challenges.

We describe a technique that takes conceptual, declarative Structure-Behavior-Function (SBF) models of a complex system, and simulates the behavior of the system in Net Logo, an agent-based simulation environment. Our technique uses a library of behavior patterns, where a behavior pattern is a parameterized generic abstraction over classes of SBF models. Given an SBF model constructed by a student, our technique first recognizes the model as an instance of a behavior pattern, then uses the parameters of the pattern to set up the Net Logo simulation, and finally runs the simulation to produce an animation that acts as feedback on the conceptual model.

Biologically inspired design uses cross-domain analogies from biology to engineering to enhance design creativity and innovation. This analogical transfer requires conceptual understanding of biological systems. In this paper, we describe a prototype interactive knowledge-based design environment called DANE for supporting conceptual understanding through learning about functional models of biological designs. We present initial results from deploying DANE in an interdisciplinary class on biologically inspired design, indicating that designers found DANE's functional models useful for conceptualizing complex systems.

Biologically inspired design is an important emerging movement in engineering design. Finding relevant biological sources of inspiration from existing biology literature is one of the important challenges of this activity. We conjecture that annotating biology articles with lightweight Structure-Behavior-Function (SBF) models is one way to address this challenge. We present Biologue, a social citation cataloging system that allows its users to gather, organize, share, and most importantly, annotate scholarly articles with SBF models. This feature not only allows the implementation of search mechanism that is more targeted to the needs of designers seeking bio-inspiration, but also helps designers make sense of the articles returned by the search mechanism.

The core problem we address in this paper is how to take a declarative conceptual representation of a complex system and produce an agent-based simulation of that model. In particular, we describe a computational technique that takes Structure-Behavior-Function models of complex systems and simulates the behavior of the modeled system in NetLogo, an agent-based programming and simulation environment. This technique has been implemented in an interactive learning environment to promote complex systems learning among middle school students.

Analogy is a fundamental process of creativity. Biologically inspired design by definition entails cross-domain analogies, and in practice has led to many innovative designs. Thus, biological inspired design is an ideal domain for studying creative analogies. In this paper, we describe an intricate episode of biologically inspired design that unfolded over an extended period of time. We then analyze the episode in terms of Why, What, How and When questions of analogy. This analysis provides a content theory of creative analogies in biologically inspired design.

The ability to adapt is a key characteristic of intelligence. In this work we investigate model-based reasoning for enabling intelligent software agents to adapt themselves as their functional requirements change incrementally. We examine the use of reflection (an agent's knowledge and reasoning about itself) to accomplish adaptation (incremental revision of an agent's capabilities). Reflection in this work is enabled by a language called TMKL (Task-Method-Knowledge Language) which supports modelling of an agent's composition and teleology. A TMKL model of an agent explicitly represents the tasks the agent addresses, the methods it applies, and the knowledge it uses. These models are used in a reasoning shell called REM (Reflective Evolutionary Mind). REM enables the execution and incremental adaptation of agents that contain TMKL models of themselves.

We view sensemaking in threat analysis as abducing a story whose plot explains the current data and makes predictions about the future. We have developed a preliminary computational system, called STAB, that abduces stories from data. STAB abduces multiple competing hypotheses by retrieving and instantiating story plots matching the current evidence. The story plots in STAB are represented as processes with goals and states, and organized in an abstraction hierarchy. STAB analyzes the VAST-2006 dataset. Given the VAST data incrementally, STAB generates multiple explanatory hypotheses, calculates their confidence values, and generates expectations about future data.

There is an increasingly large demand for software systems which are able to operate effectively in dynamic environments. In such environments, automated software engineering is extremely valuable since a system needs to evolve in order to respond to changing requirements. One way for software to evolve is for it to reflect upon a model of its own design. A key challenge in reflective evolution is credit assignment: given a model representing the design elements of a complex system, how might that system localize, identify and prioritize prospective candidates for potential modification. We describe a model-based credit assignment mechanism. We also report on an experiment on evolving the design of Mosaic 2.4, an early network browser.

Software evolution is the most costly and time-consuming software development activity, yet software engineering research is predominantly concerned with initial development. MORALE is a development method specifically designed for evolving software. It features an inquiry-based approach to eliciting change requirements, a reverse engineering technique for extracting architectural information from existing code, an approach to impact assessment that determines the extent to which the existing system's architectural components can be reused in the evolved version, a reflective approach to actually perform the evolution, and a specific technique for dealing with the difficulties that arise when evolving user interfaces. MORALE is described in the context of making a specific change to an existing system: adding user-configurable viewers to Version 2.4 of the Mosaic Web browser. Issues that arise are discussed, and the Esprit de Corps tool-suite is described

The general task of abduction is to infer a hypothesis that best explains a set of data. A typical subtask of this is to synthesize a composite hypothesis that best explains the entire data from elementary hypotheses which can explain portions of it. The synthesis subtask of abduction is computationally expensive, more so in the presence of certain types of interactions between the elementary hypotheses. In this paper, we first formulate the abduction task as a nonmonotonic constrained-optimization problem. We then consider a special version of the general abduction task that is linear and monotonic. Next, we describe a neural network based on the Hopfield model of computation for the special version of the abduction task. The connections in this network are symmetric, the energy function contains product forms, and the minimization of this function requires a network of order greater than two. We then discuss another neural architecture which is composed of functional modules that reflect the structure of the abduction task. The connections in this second-order network are asymmetric. We conclude with a discussion of how the second architecture may be extended to address the general abduction task.

Abductive inferences seem to be ubiquitous in cognition, and cognitive agents often solve complex abduction tasks very rapidly. However, abduction characterized as ‘inference to the best explanation’ is in general computationally intractable. This paper describes three related ideas for understanding how intelligent agents might efficiently perform abduction tasks. First, we recharacterize the abduction task as inference to a confident explanation, where a confident explanation is internally consistent, parsimonious, distinctly more plausible than alternative explanations, and explains as much of the data as possible with high confidence. Second, we describe a decomposition of the task of synthesizing a confident explanation into several subtasks so that the synthesis starts from islands of relative certainty and then grows opportunistically. This decomposition helps in controlling the computational cost of accommodating interactions among explanatory hypotheses, especially incompatibility interactions. Third, we present a concurrent mechanism for synthesizing confident explanations. The mechanism exploits data and processing dependencies afforded by the decomposition of the synthesis task. The emphasis of this approach to abduction is on characterizing the constraints of the abduction task and exploiting these constraints for making abductive inferences. In describing this approach, we also clarify the precise class of abduction problems addressed by the RED-2 system, and report on some new experiments. The main result is a computational model that not only enables efficient abductive inferences but also accommodates explanatory interactions, uncertainty, and data collection.

The authors have developed a computer simulator called 'NCHIPSIM' which can be used to simulate with a microcomputer the performance indicators of an integrated circuit microprocessor chip based on silicon NMOS technology. In addition to predicting the various chip performance indicators such as its size, power consumption, maximum clock frequency, computational capacity, functional throughput and the fabrication yield for a chip with given technology parameters, the simulator can also be used to simulate the dependence of any of the performance indicators based on the technology feature size as well as on the integration level of the chip.>

The authors deal with the comparison of quaternary and binary logic with reference to entropy, speed of data transmission and data string length QUATLOG, computer simulator developed, demonstrates the relative advantages of employing quaternary logic for data transmission.>

An efficient implementation procedure has been developed for the realization of sequential circuits using PLAs. The synthesis procedure is simple and based on a heuristic approach. Synchronous sequential circuits which have been widely used in digital computers over the years can be easily implemented in a single chip layout. One of the major advantages of this method is the reduction in chip area in terms of the fusible links blown to realize the state machine using PLAs.>

The authors have developed a computer-efficient algorithm and the related CAD oriented software to calculate the transverse propagation delay in a MODFET. The model has been used to study the dependence of these delays on the various MODFET parameters. The results can be utilized for the optimization of high-speed circuits.>

A set of experiments that precisely identify the power and limitations of the method of back-propagation is reported. The experiment on learning to compute the exclusive-OR function suggests that the computational efficiency of learning by the method of back-propagation depends on the initial weights in the network. The experiment on learning to play tic-tac-toe suggests that the information content of what is learned by the back-propagation method is dependent on the initial abstractions in the network. It also suggests that these abstractions are a major source of power for learning in parallel distributed processing networks. In addition, it is shown that the learning task addressed by connectionist methods, including the back-propagation method, is computationally intractable. These experimental and theoretical results strongly indicate that current connectionist methods may be too limited for the complex task of learning they seek to solve. It is proposed that the power of neural networks may be enhanced by developing task-specific connectionist methods.>

Structured matching is a task-specific technique for selecting one choice out of a small number of alternatives based on a given set of parameters. In structured matching, the knowledge and control for making a decision are integrated within a hierarchical structure. Each node in the hierarchy corresponds to a different aspect of the decision and contains knowledge for directly mapping the results of its children nodes (or selected parameters) into a choice on the subdecision. The root node selects the final choice for the decision. The authors formally characterize the task and strategy of structured matching and analyze its computational complexity. They believe that structured matching captures the essence of what makes a range of decision-making problems computationally feasible to solve.>

The general information-processing task of classification is considered and reviewed from the perspectives of the knowledge-based-reasoning, pattern-recognition, and connectionist paradigms in artificial intelligence, paying special attention to knowledge-based classificatory problem solving. The authors trace the evolution of the mechanisms for classification as the computational complexity of the problem increases, from numerical parameter-setting schemes, through those using intermediate abstractions and then relations between symbols, and finally to complex symbolic structures that explicitly incorporate domain knowledge.>