Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor...

16
Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao March 22 , 2017

Transcript of Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor...

Page 1: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Bayesian Poisson Tensor Factorization (BPTF)

for feature selection from GDELT

Chennan Zhou, Feiyan Liu, Jianbo Gao

March 22 , 2017

Page 2: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

➢ GDELT collects nearly all events about human activity in the world since 1979.

➢ How can we analyze the events we are concerned?

➢ Will BPTF proposed by David M. Blei and Hanna Wallach help us extract the

desired events?

Purpose

➢ Answer: extract the relevant events from. However, this is difficult since events

in GDELT do not contain sufficient information as which one is precisely which,

and each day hundreds of thousands events happen.

Page 3: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

➢ Discussions

Outline

➢ Introduction: Tensor Factorization

➢ The decomposition of the events from GDELT by BPTF

Page 4: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

𝑇 =

𝑖=1

𝑘

𝜋𝑖𝑢𝑖 ⊗𝑢𝑖 ⊗+𝜀𝑅.

➢Tensor analogue of matrix eigen-decomposition.

What is tensor (CP) factorization?

➢Goal: Given T with noise, 𝜀𝑅, recover factors 𝑢𝑖.

Page 5: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

How BPFT works?

➢ Input

• 4 - way tensor, whose structure is as:

<Sender, Receiver, ActionType, Date>

• Elements (𝑦𝑖𝑗𝑎𝑡) of a tensor are the number of events that sender 𝑖 acts

on receiver 𝑗 with action-type 𝑎 on date 𝑡.

➢ Output

• k components, each of which consists of four factors:

Senders, Receivers, Action-type, Time-steps.

Page 6: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Real Event: In July 2016, four consecutive violent attacks took place in Germany in a week.

Violent Attacks In Germany

Page 7: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Violent Attacks In Germany

Decomposition compare with the original time series

Page 8: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Real Event: From June to July 2016, three successive terrorist attacks took place in Turkey.

Successive Explosion In Turkey

Page 9: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Successive Explosion In Turkey

Decomposition compare with the original time series

Page 10: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Real Event: On the evening of July 14, 2016, a truck was deliberately driven into crowds in Nice,

France, resulting in the deaths of 86 people and injuring 434.

Terrorist Attack In France

Page 11: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Terrorist Attack In France

Decomposition compare with the original time series

Page 12: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

July 12, 2016 the South China Sea Arbitration

Page 13: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Examples that some of the components couldn’t be related to certain event

Page 14: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

Examples that some of the components couldn’t be related to certain event

Page 15: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

➢ When to use BPTF:

Discussions

• When we don’t know what happened in a period of time, it can extract some

events from database and obtain the evolution of events;

• It can provide guidance for the selection of features when we do not know

what fields to select from database.

➢ can not yet:

• Extract precisely event we are concerned;

• Predict the evolution of major events in the future;

• Describe complex network among countries during the evolution of a particular

event.

Page 16: Bayesian Poisson Tensor Factorization (BPTF) for feature ... · Bayesian Poisson Tensor Factorization (BPTF) for feature selection from GDELT Chennan Zhou, Feiyan Liu, Jianbo Gao

THANKS