홍석쓰 블로그

•What is the appropriate and differentiable norm when given data points with outliers? (median is not differentiable) •How do you define outliers? What is the probabilistic meaning of outliers? 첫 번째 질문의 답이 p-norm with high p였던 거 같아서 p-norm, 2-norm, frobenius norm을 찾아보다가 svd, eigenvalue decomposition을 찾아보다 왜 p-norm with high p인 지 아직도 이해를 못함. 일단 induced norm 인가 schatten norm인 지 모르겠음. https://chris..

09.18.2024Gradient Penalty라고도 불리나 봄. Applying a gradient penalty in GAN training (especially in WGAN-GP) helps stabilize the training process by enforcing the Lipschitz continuity condition. This regularizes the discriminator’s gradients, ensuring that they remain well-behaved, which in turn helps the generator receive more meaningful updates. It addresses critical issues like mode collapse, van..

Signal Processing in Computer Vision 복습. 카메라 팀에서는 이게 기본이구나. Radiometry, Stereo Depth 물어볼 줄 알았었는데. 아니었음. 이 짱짱 교수님 강의 정리해봄: https://www.youtube.com/watch?v=m_11ntjkn4k&list=PL2zRqk16wsdorCSZ5GWZQr1EMWXs2TDeu&index=8 Fourier Transform. Do you know Fourier Transform? What are the properties of Fourier Transform? If you have a periodic function, it can be expressed as a sum of sinusoids without loss ..