Actionable Intelligence from Multisourced Events

1
Actionable Intelligence from Multisourced Events

• Roy Williams
• Caltech
• with S.G.Djorgovski, C. Donalek, A. Drake, M.
  Graham, A. Mahabal, R. Seaman (NOAO).

2

• VOEvent
• VORapid
• Skyalert Infrastructure
• Decision Human Archive Machine

please pray for VAO
3
What is VOEvent

• XML document according to VO-standard
• Who, Wherewhen, Why, Params
• Each event belongs to a Stream
• key-value-semantic records
• Author, Subscriber, Broker functions
• Information not Imperative
• Follow-ups are other VOEvents
• Connected in citation graph to form Portfolio

4
Microlensing event (OGLE)
5
Global Event Authors and Followup
GCN _at_ NASA/GSFC SWIFT Fermi Integral AGILE
AAVSO
LCOGT
Liverpool Telescope La Palma
Faulkes Hawaii/Australia
PannSTARRS
eStarExeter
Caltech
UKIRT Hawaii
OGLE III Poland
NOAOTucson
Palomar P60 Caltech
Stream Author Publisher Repository Relay Followup
Subscriber
Pairitel Berkeley
CTIO/KPNO
MOA
Catalina RTS UAriz
6
VORapidVirtual Observatory Rapid Transients
Facility
please pray for VAO

• Content
• VOEvent
• Consume
• Browse, Query, Subscribe, Decision
• Author
• Streams, Alerts, Automation
• Portfolio
• Annotation, Mining
• Brokering

• Transport
• Tier 1
• Broker, Forward
• Tier 2
• Applications

7
Tier1 and Tier2 Event Nodes
please pray for VAO
Repository (Tier2)
Author (Tier2)
Jabber/XMPP or custom TCP?
Broker (Tier1)

• Tier1
• Immediate forwarding, standard capable
  protocols
• Tier2
• Subscription service, Repository, Query, Archive,
  Machine Learning, etc etc

8
International
GCN Broker
Annotation from archives
SkyAlert
Astronomers Amateurs Students
Microlensing Optical transients Radio
transients X-ray transients Gamma
transients Grav. waves Neutrinos
Events and annotation disseminated to subscribers
in real time with intelligence
Followup Scheduler
Telescope
Telescope
Telescope
Event Authors
Event Subscribers
9
Stream as Event TemplateTaming Multisourced Data

stream
event
made byrobot system at night
made byperson in daytime
10
Future .... ASKAP, IceCube LIGO/Virgo, MWA,
SkyMapper, Veritas TeV
secondary streams
11
Portfolio of One First Discovery VOEvent
12
Rich portfolio Supernova 2007sr in Antennae
13
Event-Action-Event Cycle
14
Google Sky has VOEvents
Thanks to Ryan Scranton
15
WWT Realtime Event Display
Thanks to Jonathan Fay
16
Bayesian Learning

• Feature Vectors
• Best Recommendation
• Combine with Human judgment
• Error bars, upper limits, and missing values
• All are part of the prior
• Summing opinions of multiple experts
• Some not experts!
• Relevance Vector Machine
• Best of training set (most learning)
• Tutorial
• Escalation to real expert

17
Building Feature Vector
Mahabal et al arXiv0810.4527 astro-ph
18
Recommendation for Follow-Up
Mahabal et al arXiv0802.3199 astro-ph
Sparse data ? Ambiguous classi?cation ? best
follow-up strategy to reduce confusion Example
optical light curve with a particular time
cadence would discriminate between a Supernova
and a quasar, Example particular color
measurement would discriminate between a
cataclysmic variable eruption and a gravitational
microlensing event,
19
Human Volunteers

• Science Layer
• Describe what you see in image
• Each person has level of expertise
• How to use data most effectively
• Game Layer
• Makes people come back
• Top 10 ranking etc
• Anonymous partner a la gwap.com

20
Human Computing

• Ask humans to describe
• not interpret

21
Automated Decision through Tripod of Data

• Archive
• nearby radio source escalates p(blazar)
• nearby galaxy escalates p(supernova)
• Human
• Crowded field? Artifact present?
• Can make follow-up observation
• Machine
• Fuzzy center escalates p(host galaxy)
• Moving source escalates p(asteroid)
• Bobotic follow-up observation

decision
human
archive
machinelearning
22

• Please try skyalert.org
• register, then set an alert
• Do you have a stream of astronomical events?(and
  can I have them)
• Who knows how to make a scalable push
  network?(can we talk?)