home news faculty undergraduate graduate staff seminars resources & links history intranet giving alumni

email:  taylor@biology.ucla.edu
phone:  56850
fax:  310 206-3987
office:  LS 3109
lab:  LS 3113
homepage:  http://taylor0.biology.ucla.edu

research interests:  Research interests: population genetics and adaptation; population structure of malaria vectors; adaptation in sensor arrays, especially for bioacoustics.

Research Interests

Research Interests

1. Artificial Life

Some of the most important unsolved questions in science involve issues of self-organization and emergent behaviors. They underly our understanding of adaptation, mind, and even life itself.
Good platforms for studying these problems are
distributed arrays of sensors and effectors or autonomous vehicles that are self-organized
to behave as a single ensemble.

My current research, with Ed Stabler in the Linguistics Department at UCLA,
and our students is directed at evolving and teaching sensor arrays or
and networks of autonomous vehicles to recognize concepts and to discourse intelligently about them.
In a classic essay "What is it like to be a bat?" Thomas Nagel argued that the way bats conceptualize and organize their world would be radically different than the way humans do. It is likely that the world of sensor arrays and unmanned vehicles would be even more radically different than ours.
Much could be gained if such inanimate systems could ground and evolve languages with which to discourse amongst themselves and with us.

Recent summaries of our research are described in Lee et al, 2003 and Stabler et al, 2003, in the references below. We identify three main problems that need to be addressed: (1) how to ground concepts or words to features in the flood of perceptions that come into sensors or autonomous vehicles.
though not usually approached this way, our group has been exploring how this can be viewed as a problem in learning theory and, in that light impose reasonable constraints that do permit grounding of infinite languages to infinitely complex observations ; (2) Once symbols are grounded and a language for discoursing about them is agreed upon, then then language must be learned by other agents. Simple finite languages are provably learnable, but these are typically not able to express complex ideas (e.g. the propositional calculus is larger than the class of languages that are learnable in the sense of Gold), while more complex languages (e.g. the set of all context sensitive languages) may be capable of expressing complex ideas (e.g. all of first order predicate logic) but these languages are typically provably not learnable. With Ed Stabler my students and I have been exploring how a subset of context-sensitive languages can be both sufficiently expressive and learnable. We have been using our results to explore how arrays of computing agents can learn to describe and communicate about events in their perceptual worlds. (3) Finally, we believe such arrays should to be able to evolve, so that they can acquire robust, adaptive and distributed knowledge. We are just beginning to address these challenges, but have had good success with similar problems in the past (see, e.g. Mitchell and Taylor, 1999 for references).

This work is supported principally by DARPA (with the Army and Air Force) and by the NSF. For more information about the students, publications and research projects see our informal web page, http://taylor0.biology.ucla.edu with links to the Adaptive Language Group, and the UCLA-CENS web page, http://cens.ucla.edu/.


2. Malaria and Anopheles gambiae

We currently have two principal projects relating to the population structure of Anopheles gambiae and closely related species in Africa. These are the principal vectors of malaria and of great practical importance.

2a. Population genomics of Anopheles gambiae.
After initial success in reducing the incidence of malaria the world-wide effort at malaria control has floundered, so that malaria has now been increasing, especially in Subsaharan Africa. It is estimated that over 200 million people there suffer from malaria each year, with approximately 150-200 children dying each hour. Most of the traditional means for malaria control no longer work so well as they used to, if they work at all, and the countries there lack the resources to mount effective control programs. As a result, several radically new methods for malaria control are being explored by the international health community: a malaria vaccine and efforts to genetically modify the local mosquito populations so that they no longer can transmit malaria. It is this latter approach we are helping with.

We are concerned with the population structure of Anopheles gambiae s.s., with an emphasis on its utility for guiding efforts to introduce a genetically modified vector to control malaria. This includes learning population sizes and gene flow among the populations there. We are also concerned with genetically characterizing the populations there and learning the basis for genetic differences. The population structure of An. gambiae is quite complex. The first taxonomic level in the system is Anopheles gambiae sensu lato (s.l.) that comprises 6-7 species, one of which is Anopheles gambiae sensu stricto (s.s.). It in turn has as many as 5 different ?chromosomal forms?. In some locations (e.g. Mali), the distributions of chromosomal forms coincide with ?molecular forms? which can be distinguished with polymerase chain reaction (PCR) assays. This is not, however, the case in all locations. Gene flow across the forms is limited, so these are an important part of the population structure. For a recent description of the population structure around Banambani, Mali, see Taylor et al (2001) and Taylor and Manoukis (2003).

Our efforts so far have been concerned primarily with Mali, and to some extent Burkina Faso. We have recently begun to extend our work to Cameroon and other locations in Africa.

2b. Computer model of malaria transmission in an irrigated area near Niono, Mali.

Our aim is to construct a computer simulation of malaria transmission for
the irrigated rice fields of Niono, Mali and surrounding non-irrigated villages. We
are developing and parameterize a computer model largely from samples of breeding
sites, adult mosquitoes, man biting, infection rates, survival rates, human blood
index and of parasite prevalence from humans from the surveys of villages made
during 1995-1998 by Yeya Touré and colleagues, supplemented by remote sensing images and
additional observations that we are making (see e.g. Diuk Wasser et al, 2003). We will validate the model in a
series of predict/test/refine cycles. We will then test it by comparing its
predictions of entomological indices of malaria transmission to those obtained from 18
villages in the same area during 1999-2001. We will analyze the model to obtain
mechanistic explanations about why large irrigation projects such as Niono sometimes
increase the level of malaria transmission and sometimes reduce it. Once the model is
operational we will ask a series of ``What if ??" questions to explore operational
features that are likely to increase or diminish the malaria problems accompanying
such irrigation projects; we will integrate our models with the administrative support
structure now being developed for the Niono irrigation project.

This work is being done in collaboration with the Malaria Research and Training Center (MRTC) at the National Medical School in Bamako Mali, and funded primarily by the NIH. For more information about the students, publications and research projects see our informal web page, http://taylor0.biology.ucla.edu with links to the Malaria Group.

Selected Publications

Trifa, V.M, A. Kirschel, Y. Yao, C. Taylor and L. Girod. 2008. From bird species to individual songs recognition: automated methods for localization and recognition in real habitats using wireless sensor networks Computational Bioacoustics for Assessing Biodiversity BfN-Scripten of German Federal Agency for Nature ConservationBonn 73-82 .

Trifa, V.M., A.N.G. Kirschel, and C.E. Taylor. 2008. Automated species recognition of antbirds in a Mexican rainforest using hidden Markov models Journal of Acoustical Society of America 123: 2424-2431 .

Manoukis, N. C. , J. R. Powell, M. B. Touré, A. Sacko, F. E. Edillo, M. B. Coulibaly, S. F. Traoré, C. E. Taylor and N. J. Besansky. 2008. A test of the chromosomal theory of ecotypic speciation in Anopheles gambiae Proc. Natl. Acad. Sci. USA 105: 2941-2945 .

Escobar, I., E. Vilches, E.E. Vallejo, M.L. Cody and C.E. Taylor. 2007. Self-organizing acoustic categories in sensor arrays Advances in Artificial Life: 9th European Conference, ECAL 2007 SpringerLisbon, Portugal 1161-1171 .

Trifa, V.M., L. Girod, T. Collier, D.T. Blumstein and C.E. Taylor. 2007. Automated wildlife monitoring using self-configuring sensor networks deployed in natural habitats Proceedings of the Twelfth International Symposium on Artificial Life and Robotics (AROB 12th 2007) Beppu, Japan .

Diuk-Wasser, M.A., M.B. Touré, G. Dolo, M. Bagayoko, N. Sogoba, I. Sissoko, S.F. Traoré and C.E. Taylor. 2007. Effect of rice cultivation patterns on malaria vector abundance in rice-growing villages in Mali Am. J. Trop. Med. Hyg 76: 869-874 .

Ali, A., T. Collier, L. Girod, K. Yao, C. Taylor and D.T. Blumstein. 2007. An Empirical Study of Acoustic Source Localization IPSN '07: Proceedings of the sixth international conference on Information processing in sensor networks Cambridge MA. ACM PressNew York, NY .

Lee, Y., T.C. Collier, C.E. Taylor and E.P. Stabler. 2007. Cohesion of languages in grammar networks Cooperative Control of Distributed Multiagent Systems Wiley-InterscienceChichester, U.K 359-376 .

Vilches, E., I. Escobar , E.E. Vallejo and C.E. Taylor. 2007. Targeting Input Data for Acoustic Bird Species Recognition Using Data Mining and HMMs IEEE international Conference on Data Mining: Workshop on Data Mining and Uncertainty (DUNE. Omaha Nebraska, October 28-31, 2007 .

Vallejo, E.E., M.L. Cody and C.E. Taylor. 2007. Unsupervised acoustic classification of bird species using hierarchical self-organizing maps IEEE international Conference on Data Mining: Workshop on Data Mining and Uncertainty (DUNE. Omaha Nebraska, October 28-31, 2007 ACALSpringer-Verlag Berlin 212-221 .