yuns

http://web.stanford.edu/class/cs224w/ Graph의 정의 Graphs: a general language for describing and analyzing entities with relations/interactions Network와 Graph의 종류 Networks (=Natural Graphs) Social Networks Communication and transactions Biomedicine Brain Connections Graphs Information/knowledge Software Similarity Networks Relational Structures Graphs: Machine Learning Complex domains have a rich r..

Motifs and Structural Roles in Networks SubNetworks (subgraphs) 네트워크의 부분집합 charaterize와 discriminate 가능 Case Example of SubGraphs 방향 그래프에서 나올 수 있는 모든 가능한 subgraph들 subgraph "significance"를 분류 가능한 metric 고려 Negative value: Under-representation Posivite value: Over-representation Network significance profile feature vector with values for all subgraph types Diverse domain Regulatory network(gene r..

z-score: z = $X-\mu / \sigma$로 계산하여 표본이 평균으로부터 몇 구간의 표준편차만큼 떨어져 있는지 알려주는 값 Motifs and Structural Roles in Networks SubNetworks (subgraphs) 네트워크의 부분집합 charaterize와 discriminate 가능 Case Example of SubGraphs 방향 그래프에서 나올 수 있는 모든 가능한 subgraph들 subgraph "significance"를 분류 가능한 metric 고려 Negative value: Under-representation Posivite value: Over-representation Network significance profile feature vector ..

youtube & 강의자료 Network Properties: How to Measure a Network Plan: Key Network Properties 그래프 평가 요소 Degree Distribution: $P(k)$ Path length: $h$ Clustering coefficient: $C$ Connected components: $s$ 1) Degree Distribution Degree Distribution $P(k)$: degree가 $k$개인 노드를 random하게 선택할 확률 $N_k$: # of nodes with degree $k$ Normalized histogram $P(k) = \frac{N_k}{N}$ 2) Path in a graph path: 각 node가 다음 n..

youtube&강의자료 Why Networks? Networks are a general language for describing complex systems of interacting entites Two Types of Networks/Graphs Networks Society: 70억개 이상의 individual의 collection Communication System: link electronic edevices Genes/protein의 interaction으로 생명 조절 Thoughts: 뇌의 수십억개의 뉴런의 connection으로 연결되어 있음 Information Graphs Information/Knowledge: 구성되고 연결됨 Scene Grphs: scene에서 object들이..