Highlights of Russ Greiner's CV (2007)

Update: 14/Jan/06 to present (25/Jan/2007)

Research:
- 3 journal papers -- Nucleic Acid Research, Clinical Cancer Research, (Journal of Computers)
- 9 papers in major conferences [5 Tier1*] -- IJCAI07*, NIPS06*, PKDD06 (2), COLING-ACL06, ICML06*, IAAI06, CdnAI06, ECCV06*
- (too) many posters, minor conferences, ...
- Current totals:
  - 25 Journal
  - 60 Fully-refereed Conferences
  - 46 Lightly-refereed Conferences/Workshops/Symposia
  - + 18 other -- book chapters, ...
Popular press:
- AICML won the ASTech Award; see Announcement
- I was interviewed several times, by
  - "Business Edge": Machine learning attracting major players
  - Albeta Venture: tba
  - Genome Alberta, Alberta researchers 1st to complete the human metabolome, 23 Jan 2007.
  - Erin Ottosen, to appear.
  - Innovation Alberta, Web IC: The Intuitive Browser, 6 June 2006 (#203)
Service:
- I am (again) director of AICML!
- I served as the Conference Chair of ICML 2006
- As well as usual reviewing: IJCAI, AAAI, UAI, AIJ, CI, NSERC, ...

Update: 1/Feb/05 to 13/Jan/2006

Research:
- Awarded IJCAI'05 Distinguished Paper prize
  (Also a poster won 2nd Prize in CPI'05)
- 3 journal papers -- Nucleic Acid Research, Machine Learning, Artificial Intelligence
- 11 papers in major conferences [7 Tier1]-- *AAAI05 (2), *IJCAI05 (2), *ICML05, *UM05, ECML05, PKDD05, CIBCB05, ICMLA05, *IUI06
- (too) many posters, minor conferences, ...
Popular press:
- The gov't of Alberta snail-mailed everyone this Alberta Surplus document,
  which lists only 4 main research results (p3, left).
  One is my "Brain Tumor" project!
- These results also appeared on a radio interview and in the Edmontonians journal.
Service:
- I am starting to serve as the Conference Chair of ICML 2006
- As well as usual reviewing: IJCAI, AAAI, UAI, AIJ, CI, NSF, NSERC, ...
Commercial:
- Provision patent for automated brain tumor segmentation
- We are founding a company to work on topics related to WebIC

Executive Summary (including Major Awards)

McCalla Professorship; University of Alberta; 2005-06
3 Paper Prizes
Founding and Current Director of Alberta Ingenuity Centre for Machine Learning
Program Chair for ICML'04
> 100 refereed papers
> $5M/year in funding
Editor-in-Chief of "Computational Intelligence"
Instrumental in bringing AAAI/ISMB/KDD/... to Edmonton in 2002

MOST SIGNIFICANT RESEARCH CONTRIBUTIONS

(References [Ji], [Cj], ... refer to specific papers -- journal, conference, ... -- as indexed in CV.)

1. Application pull

WebIC
a client-side Web recommender system that predicts the user's information needbased on his browsing patterns, then points him to webpages, from essentially anywhere on the Web, that contain information useful to that user.
[C7, C8 (Best Student Paper); O1, O10, O11, O12]

medical-informatics and bio-informatics results

using multiple SNPs to predict cancer [J4]
using properties of known homologues to predict properties (GO function, subcellular localization) of novel proteins [J2, J3, J5]
using NMR spectra to simulateously detect and quantify hundreds of metabolytes in complex bio-fluids (eg, urine) [J8]

my implementation is currently used by ChenomX

locating brain tumours within MRImages, then predicting their growth, to better determine appropriate radiation treatment

2. Results that use theoretical analysis to extend existing learning algorithms

Budgeted Learning
Given a fixed budget to spend collecting training data, identify which features of which training examples should be included in the training sample, to produce an optimal classifier. This decision is done sequentially, at each step based on the current results and the costs of obtaining the value of each feature.
[C1, C4; 02]

Learning Discriminative Probabilistic Models

ELR ("structural Extension of Logistic Regression") which produces better classification performance over wide range of domains
[J1; C3, C9, C17]

learning good discriminative belief net structures
[J7; C13 (RunnerUp, Best Paper), C16]

other results related to belief nets:

compute error-bars around belief net responses [C11]

model selection for generative models [C15]

PALO (stochastic hill-climbing algorithm)

perform a hill-climbing search when the "utility function", used to evaluate the system's performance, can only be estimated by sampling

published theoretical results + many applications

[eg, improving efficiency of Prolog program; robot navigation; ...]
[J13, J14; C25, C27, C28, C29, C31 (Best Paper Prize); B4, B5]

Theory Revision

incrementally improves a good, but imperfect, domain theory

published theoretical analyses [J10, J11; C21, C23]

deployed implementation based on "Delta"

2 US patents issued

Efficient Image Interpretation

Building an image interpretation system that is efficient, as well as accurate
[C2, C5, C12; O3, O4, O5, 08]

3. Theoretical results that improve on standard learning framework

Sequential learners:

using new techniques to improves on PAC-learning sample complexity bounds [C22, C24, B3]

Dealing with Missing Information

Determining which algorithms are appropriate for different

independent blocker [C26;B2]
malicious blocker
"irrelevance" blocker [J12,C20]

Building learners that can learn active classifiers,
that can actively request unspecified information [J6; C19]

4. Other minor/dated results

ACRONYM vision system, meta-level representation language (RLL, MRS),
"Novelty", Analogical inference,
Analysis of EBL, Dynamic observers (control theory), signal processing,
AndOr trees, Predicting unix commands, Pathology in search,
Focus within Reinforcement learning, ...

Bean counting:

21+3 articles in refereed journals (11 in "Artificial Intelligence")
4 edited books, special issues of journals
40 "real" conference articles
including 2 BestPapers and 1 RunnerUp
41 other (lightly refereed) publications

FUNDING

Founding Director of
Alberta Ingenuity Centre for Machine Learning
(funding at > $1.2M / year)

Currently participating in grants of > $5M / year
(of this I was instrumental in bringing in ~ $3M/year)

SERVICE

Program (co)Chair of ICML'04 [E1]

Program (co)Chair of AI&MATH 97 [E2]
Program Chair, CLNL'96 [E4]

Instrumental in bringing several major conferences to Edmonton in 2002

AAAI (1200 attendees -- first time outside US), ISMB (1600), UAI (350)
I was Local Arrangements Chair AAAI, ISMB, UAI
... and also helped on several others
These conferences were universally regarded as tremendous successes.

AIxploratorium: an on-line interactive resource for AI

(co)organized successful IJCAI'01 workshop on this topic

(co)Editor-in-Chief, Computational Intelligence, 2001-present
Revitalized this journal by changing its direction, to obtain papers in nitches not currently occupied by the other leading AI journals

On several editorial boards
(Machine Learning Journal, J of AI Research, J of MachineLearning Research, ...)

Chaired several workshops/symposia

Served on (Senior) Program Committees for many (inter)national conferences:
AAAI, IJCAI, UAI, ICML, CIRA, AI+Math, CSCSI, UserModeling, ...
+ many workshops, symposia, ...

Reviewed for (too) many journals, conferences, funders, etc:
AIJ, Comp.Opt.&Appl, IEEE PAMI, IEEE SMC, J.ACM, J.Logic Program, NIPS, FOCS, NSERC, NSF, ACM Thesis, ...

Other Activities:

Set the direction for several lines of research at Siemens Corporate Research (1991-97)

Invited talks at
- IJCAI
- Royce Conference
- > 50 at universities around the world
- > 20 public presentations to lay audiences,
  including several radio and TV presentations