### TABLE II REINFORCEMENT LEARNING PARAMETERS

2004

Cited by 2

### Table 1: Work-Based Learning Student Cases: Academic Reinforcement Findings

"... In PAGE 22: ... We also looked for examples of work-based learning positively affecting motivation towards schoolwork. (See Table1 for the results of the analysis.) Below, we give examples from our fieldwork, as well as examples, as appropriate, from the work of Moore (1981a; 1981b; 1986), and Stasz and her associates (Stasz amp; Brewer, 1998; Stasz amp; Kaganoff, 1997).... In PAGE 34: ...sometimes the experience of work in the real world has a different kind of motivational effect: two other students, Renee and Maria, had such tedious internships that they became highly motivated to attend college directly from high school, rather than delaying post-secondary enrollment or combining it with work. Summary In Table1 , we summarize the results of the analysis of our cases, noting for each student whether the three claims for academic reinforcement (school-based knowledge is applied, school-based knowledge is explored and tested, and motivation towards school is positively affected) were met. For nine of the students (over one-third of our sample), over the course of multiple visits to the internship sites, and before-and-after in-depth interviews with the students, we found no evidence for any of the claims.... ..."

### Table 3. Reinforcement Algorithm.

"... In PAGE 7: ...rom scratch. New user addition involves a similar process to reinforcement. We only need to retrieve the appropriate r vector and then add the new usern27s data via an equation similar to n2820n29. Table3 illustrates a reinforcement algorithm using all of the methods previously discussedn2c i.e.... In PAGE 7: ... Note the computational savings due to the availabilityofthe vectors r user and r imp from the original training processn2c see step n281n29. Note that the computational complexityofthealgorithm in Table3 is dominated by the Cholesky decomposition. Table 3.... ..."

### Table 1: Reinforcement learning specification for wander- ing.

"... In PAGE 4: ... We conjectured that successful wandering involves maximising the amount of carpet in view while max- imising forward velocity. Table1 shows the state, ac- tions and reward for the reinforcement learning algc- rithm. The reward is a weighted sum of the compo- nents shown.... ..."

### Table 2. Description of variables in reinforcement learning

"... In PAGE 5: ....3. Proposed parameter Using the parameter proposed in Section 2.2, the formulation in the previous section is rewritten in the frame- work of reinforcement learning as in Table2 . This problem is equivalent to the search problem in a one-dimensional gridworld, where the numbers of look-ahead steps are arranged as a discrete array.... ..."

### Table 2: Parameters used in reinforcement learning

"... In PAGE 4: ... Note that Hall and Mars found that the SLA outperformed com- mon static policies, such as FCFS, EDF and SP. We then applied our RL system under the same simulation conditions, using the parameters shown in Table2 . The mea- sured mean delay for the batch algorithm is shown in Figure 4, and for the -greedy algorithm in Figure 5.... ..."

### Table 5.3: Tracking performance of the composite controller after 3 learning trials. Results are reported for no directed exploration scheme (left), our exploration algo- rithm with initial random flailing (middle), and our exploration algorithm without initial random flailing (right).

2005