My Work & Projects

In this work, we express the masking and unique-cause MC/DC coverage criteria in Unified Combinatorial Interaction Testing (U-CIT) to test the interactions between the compile-time configuration options in highly configurable software systems. Our goal is not to propose yet another approach for achieving MC/DC coverage, but to further demonstrate the flexibility of U-CIT by applying it to a coverage criterion, which is quite different than the ones addressed by U-CIT so far. As the MC/DC criterion requires two test cases to show the independence of each condition, the test cases included in a U-CIT object cannot be generated independently from each other, which was the case for the coverage criteria addressed by U-CIT so far. Our empirical evaluations conducted on a dozen of highly configurable software systems demonstrate that U-CIT can flexibly address the aforementioned coverage criteria.

In this work, we introduce an approach, called CIT-daily, which integrates combinatorial interaction testing (CIT) with the daily build processes to systematically test the interactions between the factors/parameters affecting the system's behaviors, on a daily basis. We also develop a number of CIT-daily strategies and empirically evaluate them on highly-configurable systems. The first strategy tests the same t-way covering array every day throughout the process, achieving a t-way coverage on a daily basis by covering each possible combination of option settings for every combination of options. The other strategies, on the other hand, while guaranteeing a t-way coverage on a daily basis, aim to cover higher order interactions between the configuration options over time by varying the t-way covering arrays tested. In the experiments, we observed that the proposed approach significantly improved the effectiveness (i.e., fault revealing abilities) of the daily build processes; randomizing the coverage of higher order interactions between the configuration options while guaranteeing a base t-way coverage every day, further improved the effectiveness; and the more the higher order interactions covered during the process, the higher the fault revealing abilities tended to be.

Brain-computer interface (BCI) spellers, based on the steady-state evoked potentials (SSVEP), significantly contribute to the communication of individuals with neuromuscular disorders. These systems aim to predict a target character that a user is intended to spell as fast as possible while maintaining high accuracy. Accordingly, target character identification methods aim to reach the high information transfer rate (ITR). Methods reaching high ITR values in the literature use participants' labeled data for user calibration, which requires long and exhausting experiments for every individual that will use the speller. In this study, we developed a method that does not require labeled data from the new users; as the system is used it utilizes the accumulated unlabeled data effectively. Our method transfers the information obtained from previous users to the new user by training a deep neural network (DNN). Afterward, it uses accumulated unlabeled data of the new user to adapt the transferred DNN to that user. Adaptation is performed by assuming the DNN model's predicted target labels on the data as correct. And the model is updated in every iteration by utilizing dropout layers. Our method is compared with online template transfer canonical correlation analysis (OTT-CCA) and adaptive combined transfer canonical correlation analysis (Adaptive-C3A) methods. The comparison is performed on two large publicly available datasets (benchmark and BETA) for signal lengths between 0.2 − 1.0 seconds (s). The results have shown that our method reached approximately 5% higher maximum ITR.