Towards robust neural networks via a global and monotonically decreasing robustness training strategy

Robustness of deep neural networks(DNNs)has caused great concerns in the academic and industrial communities,especially in safety-critical domains.Instead of verifying whether the robustness property holds or not in certain neural networks,this paper focuses on training robust neural networks with respect to given perturbations.State-of-the-art training methods,interval bound propagation(IBP)and CROWN-IBP,perform well with respect to small perturbations,but their performance declines significantly in large perturbation cases,which is termed“drawdown risk”in this paper.Specifically,drawdown risk refers to the phenomenon that IBPfamily training methods cannot provide expected robust neural networks in larger perturbation cases,as in smaller perturbation cases.To alleviate the unexpected drawdown risk,we propose a global and monotonically decreasing robustness training strategy that takes multiple perturbations into account during each training epoch(global robustness training),and the corresponding robustness losses are combined with monotonically decreasing weights(monotonically decreasing robustness training).With experimental demonstrations,our presented strategy maintains performance on small perturbations and the drawdown risk on large perturbations is alleviated to a great extent.It is also noteworthy that our training method achieves higher model accuracy than the original training methods,which means that our presented training strategy gives more balanced consideration to robustness and accuracy.

Evaluation of a multidisciplinary global health online course in Mexico

Background:Global Health Education(GHE)focuses on training proactive global citizens to tackle health challenges in an increasingly interconnected and interdependent world.Studies show that health professionals in training have reported that GHE has improved their teamwork,responsiveness to contextual factors that impact health,and understanding of health systems;however,there is little research on the impact of GHE courses in undergraduate settings,especially in low and middle-income countries(LMICs).Methods:Our study analyzes a multidisciplinary online global health course at Tecnologico de Monterrey,México.We conducted a cross-sectional study with pre-and post-design.Students who took the multidisciplinary course of Global Health for Leaders in the Fall of 2019(n=153)and Spring of 2020(n=348)were selected for this study.Using a five-point Likert scale(strongly agree to strongly disagree),the survey assessed seven competencies as well as questions about course expectations,takeaways,and recommendations to improve the course.We performed descriptive statistical analyses comparing the combined pre-tests(from Fall and Spring cohorts)to the combined post-tests.Fisher’s exact test was used to compare the samples.Results:Of the 501 pre-course surveys administered,456 responses were completed in the pre-course and 435 in the post-course(91%overall response rate).Only 8.7%of the respondents in the pre-course survey strongly agreed that they could describe fundamental aspects of global health such as the Millennium Development Goals or Sustainable Development Goals,in contrast to a 56%of the students who strongly agreed in the post-course survey(p<0.001).Similar differences were captured in understanding the global burden of disease,social determinants of health,the effects of globalization in health,health systems’goals and functions,and human rights.38%felt that the course helped them develop a more empathetic perception of the suffering of others experiencing global health-related issues.Conclusion:In this study,we have presented our experience in teaching an online global health course for multidisciplinary undergraduates in a LMIC.The competencies reported by our students indicate that the course prepared them to confront complex global health issues.