Ujava.org tensor-analysis
4 likes912 views
This document provides an overview of tensor analysis concepts for deep learning. It discusses key tensor concepts like vectors, matrices, tensors, derivatives, gradients, divergence, and curl. It also covers tensor operations and their use in deep learning techniques like gradient descent and the conjugate gradient method. The document is presented by Intelligent City Ltd and their CEO Shindong Kang to provide background on tensor analysis.
Ujava.org tensor-analysis
- 1. Tensor Analysis for Deep Learning ujava.org Workshop 2015-05-03 www.idosi.com CEO 강신동 Shindong KANG (주)지능도시
- 6. www.idosi.comOsculating Circle (접촉원) P P1 P2 Lim P1P → 0 Lim P2P → 0
- 13. www.idosi.comVector ? (벡터) (1, 2) Is This Vector ?
- 14. www.idosi.comTensor ? (텐서) 11, 12 21, 22 Is This Tensor ?
- 15. www.idosi.comVector (벡터) Two vectors addition ?
- 23. www.idosi.comGradient (그라디언트, 기울기) Vector
- 26. www.idosi.comSource and Sink of Phase Plane
- 28. www.idosi.comMatrix Determinant (행렬식) Img src=wikipedia.org
- 29. www.idosi.comJacobian Matrix (자코비안 행렬) Multi variables Multi functions
- 30. www.idosi.comJacobian (Determinant) (자코비안 행렬식) = 0 변수들이 종속적임
- 32. www.idosi.comLaplacian (라플라시안, Laplace Operator) Scalar
- 33. www.idosi.comSecond-order Partial Derivative (2차 편미분) L = f(t) V = dL/dt
- 34. www.idosi.comSecond-order Partial Derivative (2차 편미분) V = dL/dt
- 35. www.idosi.comSecond-order Partial Derivative (2차 편미분)
- 37. www.idosi.comDerivative of Matrix Equation (행렬 방정식의 미분) Hessian Matrix
- 39. www.idosi.comHessian Matrix (헤시안 행렬)
- 40. www.idosi.comHessian Determinant (헤시안 행렬식) 최대값 최소값 Gradient=0
- 41. www.idosi.comConjugate Vector (켤레벡터, 공액벡터) (x,y) (x,-y)
- 42. www.idosi.comConjugate Gradient Method (켤레기울기법, 공액기울기법)
- 47. www.idosi.comDivergence (다이버전스, 발산) Scalar
- 54. www.idosi.comCurl (컬, 회전) Vector's component
- 62. Thank you ! (주)지능도시 Intelligent City Ltd. 강신동 Shindong KANG www.idosi.com ceo@idosi.com