Freshman 15: Fact or fiction?

Objective: The objective of this study was to investigate changes in body weight, BMI, body composition, and fat distribution among freshman women during their 1st year of college. Research Methods and Procedures: Freshman women during the 2004 to 2005 academic year were recruited to participate. The initial baseline visit occurred within the first 6 weeks of the fall 2004 semester, with the follow-up visit occurring during the last 6 weeks of the spring 2005 semester. At each visit, height, weight, BMI, waist and hip circumferences, and body composition (by DXA) were obtained. Results: One hundred thirty-seven participants completed both the fall and spring visits. Significant (p < 0.0001) increases between the fall and spring visits were observed for body weight (58.6 vs. 59.6 kg), BMI (21.9 vs. 22.3), percentage body fat (28.9 vs. 29.7), total fat mass (16.9 vs. 17.7 kg), fat-free mass (38.1 vs. 38.4 kg), waist circumference (69.4 vs. 70.3 cm), and hip circumference (97.4 vs. 98.6 cm), with no significant difference observed in the waist-to-hip ratio (0.71 vs. 0.71; p = 0.78). Discussion: Although statistically significant, changes in body weight, body composition, and fat mass were modest for women during their freshman year of college. These results do not support the purported freshman 15 weight gain publicized in the popular media.

Incremental learning for interactive e-mail filtering

In this article, we propose a framework, namely, Prediction-Learning-Distillation (PLD) for interactive document classification and distilling misclassified documents. Whenever a user points out misclassified documents, the PLD learns from the mistakes and identifies the same mistakes from all other classified documents. The PLD then enforces this learning for future classifications. If the classifier fails to accept relevant documents or reject irrelevant documents on certain categories, then PLD will assign those documents as new positive/negative training instances. The classifier can then strengthen its weakness by learning from these new training instances. Our experiments’ results have demonstrated that the proposed algorithm can learn from user-identified misclassified documents, and then distil the rest successfully.