02805nas a2200265 a 450000500170000000800410001703500150005810000280007324500820010126000670018330000220025049000290027250001850030150200580048652015540054465000340209865000210213265000390215370000410219270000470223370000480228071000770232881000610240585600730246620260817173419.0240229s2022    th u  m  tt   000 a eng d  a.b124216620 aChanapa Pananookooln   10aComparing selective masking methods for depression detection in social media   aPathum Thani, Thailand :bAsian Institute of Technology,c2022  a47 leaves :bill.1 aThesis ;vno. DSAI-22-04  aA thesis submitted in partial fulfillment of the requirements for the  degree of Master of Science in Data Science and Artificial Intelligence, School of Engineering and Technology  aThesis (M. Sc.) - Asian Institute of Technology, 2022  aIdentifying those at risk for depression is a crucial issue in which social media provides an excellent platform for examining the linguistic patterns of depressed individuals. A significant challenge in a depression classification problem is ensuring that the predic tion model is not overly dependent on keywords, such that it fails to predict when key words are unavailable. One promising approach is masking, i.e., by masking important words selectively and asking the model to predict the masked words, the model is forced to learn the context rather than the keywords. This study evaluates seven masking tech niques, such as random masking, log-odds ratio, and the use of attention scores. In ad dition, whether to predict the masked words during pretraining or fine-tuning phase was also examined. Last, six class imbalance ratios were compared to determine the robust ness of the masked selection methods. Key findings demonstrated that selective masking generally outperforms random masking in terms of classification accuracy. In addition, the most accurate and robust models were identified. Our research also indicated that re constructing the masked words during the pre-training phase is more advantageous than during the fine-tuning phase. Further discussion and implications were made. This is the first study to comprehensively compare masking selection methods, which has broad implications for the field of depression classification and the general NLP. Our code can be found in: https://github.com/chanapapan/Depression-Detection 0aSocial mediaxData processing 0aMachine learning 0aNeural networks (Computer science)0 aChaklam Silpasuwanchai,eChairperson1 aDailey, Matthew N.,eExamination Committee0 aMongkol Ekpanyapong,eExamination Committee2 aHis Majesty the King{u2019}s Scholarships (Thailand),eScholarship Donor2 aAsian Institute of Technology.tThesis ;vno. DSAI-22-04403Full-Textuhttp://203.159.5.9/ait-thesis/Viewer/viewer.php?id=B20417