PERSONAL DETAILS

Date of Birth

11th February 1964

Nationality

British

Contact Details

ATR Interpreting Telecommuniactions Research Labs.,

2-2-2, Hikaridai,

Seika-cho,

Soraku-gun,

Kyoto, 619-02.

Japan.

(+81) 774 95 1344

EDUCATION

1991 - 1995

University of York, Heslington, York.

1989 - 1990

University of Warwick, Coventry, Warwickshire.

1981 - 1984

University of Warwick, Coventry, Warwickshire.

1972 - 1981

Read School, Drax, North Yorkshire.

QUALIFICATIONS

D.Phil. (June 1995)

My D.Phil thesis was entitled: Density Estimation Using Neural Computer Networks examines the role of neural networks in Science statistical pattern recognition, and presents a novel dimension-reduction technique for density estimation. During the progress of my D.Phil. I gained a broad overall familiarity with neural networks from the point of view of statistical pattern recognition. There was also a significant overlap with the fields of pattern recognition and density estimation. The software used for the research were all written in C/C++ and were run on either UNIX workstations or IBM PCs. Most of the intermediate reports were written using the LATEX document preparation package. The thesis itself was prepared on an IBM PC using Microsoft Word for Windows and several other PC graphing/drawing and statistical packages.

M.Sc. (June 1990)

The course was called Cognition, Computing & Psychology. It was a multidiscipline cognitive science course covering cognitive psychology, artificial intelligence, philosophy and neural networks. This course gave me a broad view of the field of cognitive science and introduced me to the area of neural networks whilst placing it in a strong psychological context. The psychological side of the course was mainly cognitive psychology with courses in memory and psycholinguistics. The course also included a strong A.I. component and we received tuition and assignments in both LISP, PROLOG and expert system shell programming.

B.Sc. (Hons) Class 2(i). (June 1984)

This course provided me with a good Mathematics overall grounding in pure mathematics at one of the strongest universities in the country for mathematics. The course was very flexible, allowing me to take several options in computer science during in the course.

'A' Levels (June 1981)

Biology (A) Chemistry (A) Maths (A) Physics (B)

'O' Levels (June 1979)

Biology (A) Chemistry (A) Eng. Lit. (B) Eng. Lang. (C) Human Biol. (A) Maths (A) Physics (B) Geography (A) History (B) French (C)

EMPLOYMENT

2002 -Present

ATR Research Labs, Kyoto, Japan. Present This work is similar in scope to the previous contract with ATR, but also includes research in statistical machine translation between Japanese and English

2001

Tachnae Ltd., New York. This work involved the development of statistical parsing and tagging software and intitial pilot studies into their intergration into financial models of commodity price.

1997 - 2001

ATR Research Labs, Kyoto, Japan. Probabilistic tagging and parsing. In this position I have been involved in the development of a statistically based natural language tagger and parser of general English. The parser developed by my team at ATR is a state of the art decision-tree parser that uses a mult-stack decoder search algorithm. The parser is capable of parsing Programming work is in C++ on UNIX workstations.

1993-1997

Department of Computer Science, University of York. Graph matching using probabilistic relaxation labelling. This job was a postdoctoral research position concerned with object matching by means of matching graph structures derived from line segments extracted from image data. Although the work primarily concerned itself with intermediate level vision I worked on the lower level image processing tasks required to produce the objects to be matched. The computational experimentation was undertaken in C/C++ on UNIX workstations. Documentation was produced using LATEX and UNIX tools. The research was done in collaboration with the DRA at Malvern.

1986-1989

Aregon International Ltd., Vauxhall, London. Financial software development. This job involved all aspects of software development, from the design stages though to on-site support. The job provided me with a wide experience of the software development industry and involved me in a large variety of differing projects on several hardware platforms, working for major international clients such as the Amsterdam Stock Exchange, Sumitomo Bank and Credit Lyonnais. Many of the projects required me to work abroad for extended periods. The main language used was C on both VAX/VMS and UNIX platforms.

1984-1986

EASAMS Ltd., Frimley, Surrey. Flight simulation. I was a development softare engineer on two large military projects. The main language used was VAX FORTRAN.

PUBLICATIONS

1. Finch A.M. and Austin J.A., A Neural Network for Dimension Reduction and its Application to Image Segmentation, International Conference on Artificial Neural Networks, Sorrento, 1994.

2. Finch A.M. and Hancock E.R., A Bayesian Framework for Matching Delaunay Graphs, 8th International Conference on Image Analysis and Processing, 1995.

3. Finch A.M., Wilson R. and Hancock E.R., Matching Delaunay Triangulations by Probabilistic Relaxation, 6th International Conference on Computer Analysis of Images and Patterns, 1995.

4. Finch A.M. and Hancock E.R., Matching Deformed Delaunay Triangulations, International Symposium on Computer Vision, 1995.

5. Moss S., Finch A.M. and Hancock E.R., Detecting and Matching Hedges in Partial Radar Images, IAPR/TC-7 Workshop on Methods for extracting roads, buildings, roads and other man-made structures from images, August 1996.

6. Finch A.M., Hancock E.R, and Wilson R., Relational Matching with Mean Field Annealing, ICPR96, 1996.

7. Finch A.M. and Hancock E.R., Softening Discrete Relaxation, Neural Information Processing Systems, December 1996.

8. Finch A.M., Wilson, R. and Hancock E.R., Matching Delaunay Graphs, Pattern Recognition, vol. 30, pp. 123-140, 1997. (Highly Commended Paper).

9. Finch A.M., Wilson, R. and Hancock E.R., An Energy Function and Continuous Edit Process for Graph Matching, Neural Computation, vol. 10, issue 7, pp. 1873-1894, October 1998.

10. E. Black, A. Finch, H. Kashioka, Trigger-Pair Predictors in Parsing and Tagging, In Proceedings, ACL/COLING-98, Montreal, 1998.

11. R. Zhang, E. Black, A. Finch. Using Detailed Linguistic Structure In Language Modelling, Eurospeech 99, Budapest, 1999.

12. E. Black, A. Finch, R. Zhang, Applying Extrasentential Context To Maximum Entropy Based Tagging With A Large Semantic And Syntactic Tagset, Seventh Workshop on Very Large Corpora, Maryland, US, 1999.

13. R. Zhang, E. Black and A. Finch, Integrating Detailed Information into a Language Model, ICASSP, 2000.

14. R. Zhang, E. Black, A. Finch, and Y. Sagisaka, A Tagger-Aided Language Model With a Stack Decoder, ICSLP, 2000.

15. A. Finch and E. Black, Using a Parser to Aid Treebank Production, ICCPO, 2001.

16. A. Finch and E. Black, Beyond Tag Trigrams: New Local Features for Tagging, LREC, 2002.

17. A. Finch, T. Watanabe and E. Sumita, Paraphrasing by Statisitcal Machine Translation, FIT2002, Tokyo, Japan. (to APPEAR).

PATENTS

1. 2002 Sentence Paraphrasing Device Using Statisitcal Machine Translation Method, (Currently under application, With Sumata, Watanabe).

2. 2001 Trigger Pair Enhanced Natural Language Parsing Models. (Japanese Patent 3126952. With Black, Kashioka.)

3. 1999 Language Models for Speech Recognition Enhanced Via Complex Natural Language Triggers. (With Black, Zhang).

OTHER INTERESTS AND SKILLS

I am a keen portrait photographer and hold the LRPS professional qualification of the Royal Photographic Society. I like to keep myself fit and I enjoy playing squash and badminton. I have been awarded university club colours for badminton and I have played for the University of York badminton team. I enjoy working abroad, especially in Japan, and I have been studying spoken and written Japanese for 4 1/2 years and have passed the level 3 Japanese proficiency examination. I have a full, clean driving licence.