CTM, core of TBR

CTM methods are the methods which explores the useful informations in trace for reasoning, measuring the user performance, defining user profil.

One of many reasons, we worked on the CTM methods instead of applying the general ML algorithms on the interaction traces because the interaction trace is filled with an infinite number of informations that might be useful but the assistance systemm don't have the resources to pay attention to everything in interaction traces.

Stimulus

A stimulus (plural stimuli) is a detectable change in the internal or external environment.

Selection

Thus, the first step of CTM is the decision of what to attend to. Depending on the environment, and depending on the user and assistance goal, the assistance system might focus on a familiar stimulus or something new. When the assistance system attend to one specific thing in interaction trace - it becomes the attended stimulus.

Selection in CTM

The difference between interaction trace and image set is that in interaction trace, the assistance system could know about the action which changed the state of environment.

Interaction trace = actions + states.

There are 3 kinds of observations in interaction trace:

• action without state change
• action with current states
• action with effects

Organization

Once we have chosen to attend to a stimulus in the interaction trace. The choice sets off a series of reactions in our assistance, where we construct a mental representation of the stimulus (or, in most cases, the multiple related stimuli) called a percept. An ambiguous stimulus may be translated into multiple percepts, experienced randomly, one at a time, in what is called "multistable perception".

Organisation in CTM (Structure Learning)

From data to structure model. The structure model is a kind of group the observations based on the proximity and the similarity.

Percept

A percept is the input that an intelligent agent is perceiving at any given moment.[1] It is essentially the same concept as a percept in psychology, except that it is being perceived not by the brain but by the agent. A percept is detected by a sensor, often a camera, processed accordingly, and acted upon by an actuator. Each percept is added to a percept sequence, which is a complete history of each percept ever detected. An intelligent agent chooses how to act not only based on the current percept, but the percept sequence. The next action is chosen by the agent function, which maps every percept to an action.

For example, if a camera were to record a gesture, the agent would process the percepts, calculate the corresponding spatial vectors, examine its percept history, and use the agent program (the application of the agent function) to act accordingly.

Interpretation

After the assistance system have attended to a stimulus, property value pai

and our brains have received and organized the information, we interpret it in a way that makes sense using our existing information about the world. Interpretation simply means that we take the information that we have sensed and organized and turn it into something that we can categorize. By putting different stimuli into categories, the assistance system can better understand and react to the new interactions.

Interpretation in CTM (vase or 2 faces)

In human perception, attended stimulis could be categorized in the different groups at the same time. However, when labeling interaction traces, the user is quite easy to distinguish between the actions and the states.