Differences

This shows you the differences between two versions of the page.

--- public:grl_readingmemo [2024/04/04 06:10] – created liang
+++ public:grl_readingmemo [2024/04/04 07:12] (current) – liang
@@ Line 12: / Line 12: @@
 ===== #01, liangz =====
-range: from the title page to p8 (end of Chapter 1)
+range: from the title page to p8 (end of Chapter 1).
+** NOTICE: ** please read both the scanned pictures and this memo.
 i: title page
@@ Line 38: / Line 40: @@
 p1: Chapter 1 Introduction
-  * node, edge, graph => ask the audience to illustrate some graphs
+  * node, edge, relations, graph => ask the audience to illustrate some graphs
-  * Zachary Karate Club Network (1977)
+  * Zachary Karate Club Network (1977) and on its importance
+p2:
+  * It mentioned "a dramatic increase in the quantity and quality of graph data in the last 25 years." => Why 25 years? (hint: It means since 1995)
+  * ML is not the only way but may be interesting.
+  * adjacency matrix, adjacency list, simple graph and {0,1}
+  * (optional) graph processing from adjacency list to adjacency matrix seems an evidence showing human consumes matter and energy to increase its order.
+p3:
+  * multi-graph (variable number of types/relations) vs multi-relational graph (fixed number of types/relations)
+  * Heterogeneous graph (inner edges important than inter edges), multipartite graph (no inner edges, only inter edges), multiplex graph (inter edges important than inner edges)
+  * attribute or feature
+p4:
+  * graph (abstract structure) and network (real-world data)
+  * supervised (predict an output) and unsupervised (infer pattern)
+  * node classification: predict the label of a node given a small number of labelled nodes (|V_train| << |V|)
+p5:
+  * applications: bot detection in a social network, function of proteins in the interactome, classify the topic based on links, etc
+  * difference from a standard supervised learning: the assumption/bias of iid (independent and identically distributed) or no.
+  * popular inductive bias used in graph learning: homophily (same attrubute with neighbors), structural equivalence (similar local structure -> similar label), heterophily (e.g., gender).
+p6:
+  * supervised learning and semi-supervised learning, and GL (no iid assumption)
+  * relation prediction: e.g., recommendation system, side-effect. Notice the requirement of inductive bias.
+p7:
+  * clustering and community detection
+  * graph classification, regression, and clustering (to the audience: what is the general difference?)
+p8:
+  * iid assumption and why? -> Li-Yang
+  * Additional comment: Causal relation and correlation. ML is often consider the latter approach but actually we need to consider the former.