In Both Sets Of Experiments, All Participants Experienced

1303 WordsMay 7, 20176 Pages
In both sets of experiments, all participants experienced the exact same set of trials, following the same parameters. This included randomizing the order of the prescribed treatments. In this case, the treatments consisted of 200 image pairs of possible paths with obstructions of different shapes and sizes in a two-dimensional array for the participants to view and rate the relevance of the obstructions. This randomization technique is one way to minimize any carryover effect from one level to the next, and is commonly used in Within-Subjects Designs experiments (Elmes et al., 1992, p. 126). The researchers decided to use a secondary experiment with 52 participants to ensure that the results of the primary experiment were not skewed as a…show more content…
SVMs are standard tools that are used in machine learning, such as in robotics and manufacturing, and are used to recognize patterns in data and classify new data (Tanner & Itti, 2017, p. 174). The SVM algorithms were trained and then tested on the same image pairs that were used by the human experiments, and the results were compared with the mathematical goal relevance model. Descriptive Statistics The experiments provided descriptive statistics for each of the tests that were conducted. For the human experiments, whereby the participants were required to give a response based on choosing a left or right path around an obstruction, the researchers calculated inter-subject agreement for an image pair as the fraction of participants who agreed with the majority decision. The statistical data was presented graphically as a frequency distribution histogram with Number of Pairs on the y-axis and % Human Agreement on the x-axis, and numerically showed an average inter-subject agreement across all images to be 77.05% (Tanner & Itti, 2017, p. 175). The SVM computer modeling had descriptive statistics that showed the accuracy of the model in predicting the data. The statistical data was presented graphically as a frequency distribution histogram with Prediction Accuracy of Model & Inter-subject Agreement on the y-axis and % Human Agreement on the x-axis and numerically depicted a mean accuracy of 81.97%, with a standard deviation of 3.46% (Tanner
Open Document