MINDS Seminar Series | Jay Min Lee (POSTECH) - Anomaly Pattern Extraction Technique and Number Theory Vertex Algebra Interpretation Appearing in Linear Algebra Matrix Calculation: Cognitive Image Processing
period : 2021-11-30 ~ 2021-11-30
time : 17:00:00 ~ 18:00:00
개최 장소 : Online streaming(Zoom) and Math Bldg Rm 104
Topic : Anomaly Pattern Extraction Technique and Number Theory Vertex Algebra Interpretation Appearing in Linear Algebra Matrix Calculation: Cognitive Image Processing
|Date||2021-11-30 ~ 2021-11-30||Time||17:00:00 ~ 18:00:00|
|Speaker||Jay Min Lee||Affiliation||POSTECH|
|Place||Online streaming(Zoom) and Math Bldg Rm 104||Streaming link||ID : 688 896 1076 / PW : 54321|
|Topic||Anomaly Pattern Extraction Technique and Number Theory Vertex Algebra Interpretation Appearing in Linear Algebra Matrix Calculation: Cognitive Image Processing|
|Contents||If each component of a linear matrix is taken as a complex number (real number, imaginary number), there is no problem in calculating the inverse matrix because it is a nonsingular matrix. On the other hand, practical data science applications introduce the mathematical difficulty of solving the inverse problem of a singular matrix given only in real values. We established a semantically accessible computational experimental technique by directly judging a visual matrix image (each matrix component taken from a color RGB pixel image) without going through the computer's 64-bit hash function ambiguity. Such an attempt can be viewed as a technology based on applied mathematical numerical analysis because it can approach the tuning response sensitive to artificial intelligence (AI) computer vision technology from a mathematical point of view. This can be extended not only to the field of cognitive image processing but also to various artificial intelligence informatics.
A computational projection methodology through a quadrant extension matrix structure was attempted based on a mathematical topological perspective that expresses the conservation and continuity of manifolds of all invariant aspects in Hilbert space. A method of additionally generating an affine sinogram component while reducing the projection of an orthogonal matrix to a low-rank principal component matrix is presented. And in the remaining terms, we were able to solve the solution to extract the bitmap corresponding to the checkerboard notation. We are looking for anomality and affinity linkages by analyzing regularization variables in the phase image representation with the Tikhonov regularization method and the L2 norm precondition imposed. Mathematical hypotheses and insights into crypto space (eg, supersymmetric string structures in physics) are finding suitable number theories as representations of vertex algebraic functions. By inferring the repetitive numerical attraction process, it was possible to explore the reality of the cognitive system that mimics the mysterious process of the driving force of the natural computation of the universe with simplicity and endomorphism (establishing self-assembly).
MinDS · 2021-09-24 10:12 · Views 391