Slide 6
Slide 6 text
• ある時系列データが原因となり,他の時系列データに何らかの影響
を与えている現象
• 因果探索
– 独立成分分析を応用したLinGAM [Shimizu+, JMLR 2006]
• 因果指標
– Granger Causality [Granger, 1969]
– Transfer Entropy [Schreiber, 2000]
– Time Delayed Mutual Information
因果指標・因果探索
This morning, we woke to more
sobering news about the state of
our economy. The 240,000 jobs
lost in October marks the 10th
consecutive month that our
economy has shed jobs. In total,
we've lost nearly 1.2 million jobs
this year, and more than 10 million
Americans are now unemployed.
Tens of millions of families are
struggling to figure out how to pay
the bills and stay in their homes.
Their stories are an urgent
reminder that we are facing the
greatest economic challenge of
our lifetime, and we must act
swiftly to resolve them.
因果関係
Tens of millions of
families are struggling to
figure out how to pay the
bills and stay in their
homes. Their stories are
an urgent reminder that
we are facing the greatest
economic challenge of
our lifetime, and we must
act swiftly to resolve
them.
This morning, we woke to
more sobering news
about the state of our
economy. The 240,000
jobs lost in October marks
the 10th consecutive
month that our economy
has shed jobs. In total,
we've lost nearly 1.2
million jobs this year, and
more than 10 million
Americans are now
unemployed.
TEに確率的ARモデルを導入する
とGCと一致 [Shibuya+, KDD 2009]
データが多次元ガウス分布に従う
とするとTEはPartial CCA、MIは
CCAに帰着 [Shibuya+ PRE 2011]
[https://scontent-nrt1-1.xx.fbcdn.net/v/t1.0-
9/17903454_10158949965035725_3181251005684687258_n.jpg]