Deep Learning

• 프로그래밍 언어별 딥러닝 라이브러리 정리
• http://aikorea.org/blog/dl-libraries/
• 딥 러닝에 대한 실제 구현/개발 상의 조언
• https://nmarkou.blogspot.kr/2017/02/the-black-magic-of-deep-learning-tips.html?imm_mid=0eeb7e&cmp=em-data-na-na-newsltr_20170315
• 딥 러닝 관련 참조 문헌
• http://bit.ly/dlfndprep

• 참고(Tips)
• ReLU
• Weight 초기화 할 때, 분산을 2/n으로 초기화하도록. (Ref. He et al.)
• bias를 단순히 0으로 초기화하도록.
• Regularization
• L2 regularization can be expected to give superior performance over L1.
• neurons with L1 regularization end up using only a sparse subset of their most important inputs and become nearly invariant to the “noisy” inputs.