An automatic route planner for a vehicle needs to take into account distance and traffic conditions. If a main road is usually backed up at a certain time of day, it would be wise to travel an alternate route. Likewise, if a main road allows a higher speed limit, it could be faster even though the road might be longer.
Cars are fast becoming equipped with maps and route planners. A planner that only shows the shortest route might not return the fastest route. In addition, road repairs and current weather conditions can't be planned into a map, so they have to be put into the planner as data in addition to the source and destination. In addition, the driver may have reason to avoid certain routes, such as vehicle size, speed preference, or any other unpredictable reason, so the planner must be able to adjust to the driver/vehicle itself
Visual programming languages commonly suffer from scalability problems. Many attempts have been made to reduce these problems. These efforts have included the use of multiple views [1,2] and the use of fish-eyeing [3,4,5]. Each of these addresses only limited aspects of the scalability problem. The use of multiple views makes contextual understanding difficult to achieve, while the distortion or fish-eyeing approach introduces comprehension difficulties. Research in the area of comprehensible distortions may provide some relief for the fish-eye case. A carefully designed experiment may provide insight into how these distortions affect comprehension of program diagrams.
Procedural approach and declarative approaches are the two popular approaches to designing software agent communication languages. Believe that an object oriented (O-O) approach is a viable alternative. O-O objects communicate via messages which satisfies one agent communication language prerequisite. An agent communication language's subsystems,(i.e., Vocabulary, Knowledge Interchange Format, Knowledge Query and Manipulation Language), can be implemented as objects.
To determine the feasibility of using an O-O approach. I plan to investigate: (A) Vocabulary, Knowledge Interchange Format, Knowledge Query and Manipulation Language characteristics. (B) procedural and declarative approaches that are presently being used to implement agent communication languages. (C) suitable O-O alternatives.
Educational software often faces a dilemma: While the software offers a great means to teach students, often times it never gets used successfully because there is a lack of motivation on the part of the student to spend the time it takes to learn the system. Agentsheets, a domain-oriented visual programming environment, is not immune to such problems. Often times, students are scared away from using the software due to their perception of how long it will take to learn to use it. This project proposes to study and develop an interface to keep motivation levels high and anxiety low among students learning to use Agentsheets. A way to do this is to introduce new ideas to users along with concepts that they already understand, therefore easing the learning process and preventing overloading them with too many concepts at once. By taking first-time users through an example project in Agentsheets that will teach them the basics of the software, we hope to find a way to motivate students to use the software and to also speed up the learning project.
The goal of this project will be to combine a natural language input interface with a rule-based simulation of a person in an "adventure game" environment.
In our world, the title character will be able to move around and interact with his surroundings. We will implement a rule base for his actions, and the program will offer advice on how to proceed.
The implementation we are going to do as our project should, when finished, be able to classify digital images (photographs) in classes like portraits (persons), landscapes, buildings.
Systems like the one mentioned are useful for example for image databases, where each image can be accessed using characteristics of its contents (colors, persons, etc.). An example of such a system is QBIC [1]. Furthermore a system like ours would be useful for a more elaborate system like a "mixture of experts". For such a system our network could provide a rough classification, so that an appropriate subsystem is chosen to further process the image. For example, if our network says, that an image is a portrait, a subsystem would be used that can classify portraits into the classes male or female. Another objective of our project is to evaluate the classification performance of the different neural network types, as they are applied to image characterization.
Identifying interesting patterns in DNA sequence is a difficult problem. Biology often dictates that these patterns exhibit a conserved or approximately conserved sequences in order to be recognized by other cellular factors. This fact motivates searches for conserved sequence elements in genomic (DNA) sequence.
WebQuest is an interactive gaming-environment that incorporates the World Wide Web. It was developed for k-12 students to learn Internet research skills and the "theme" (topic) around which they construct or play a game. WebQuest games are constructed by the students, thereby enabling "learning by acquisition ? design, construction, and reflection." The students constructing the games become "producers of knowledge," (Perrone et al., 1996).
WebQuest was inspired by MUD/MOO/MUSH environments that developed into "entertaining" learning environments. It operates in Agentsheets and activates Netscape Navigator. It is comprised of a gameboard that is created by the students in agentsheets. After students have researched a "theme" on the Web, they formulate a list of questions and hints for their players and place them into the gameboard.
In an effort to help students design their games, we propose to embed critics into the gaming environment. We propose to construct gameboard design critics (i.e., is there a connection between land masses and paths? Are the gameboard objects being used correctly?), and critics to help students formulate questions.
In this project, I want to implement an AI program on a distributed network environment. The programming language that I want to use is MIT Scheme which may have some concurrent versions. There are a lot of studies on concurrent Scheme but I have not seen or had any version of this program to implement my code distributedly. So I will try to simulate the distributed processing in AI with regular MIT Scheme programming language.
The program that I want to implement is a "code-encryption" program. It is simply a program to decode the messages encoded with some way. The program will learn to solve this puzzle by using the data entered.(relation between frequency of letters in English alphabet and frequency of letters in encrypted code will be examined)