Manuscript symbols can be stored, recognized and retrieved from an entropic digital memory that is associative and distributed but yet declarative; memory retrieval is a constructive operation, memory cues to objects not contained in the memory are rejected directly without search, and memory operations can be performed through parallel computations. Manuscript symbols, both letters and numerals, are represented in Associative Memory Registers that have an associated entropy. The memory recognition operation obeys an entropy trade-off between precision and recall, and the entropy level impacts on the quality of the objects recovered through the memory retrieval operation. The present proposal is contrasted in several dimensions with neural networks models of associative memory. We discuss the operational characteristics of the entropic associative memory for retrieving objects with both complete and incomplete information, such as severe occlusions. The experiments reported in this paper add evidence on the potential of this framework for developing practical applications and computational models of natural memory.

No abstract

No abstract

Postal 70-275, Ciu dad Univ e rsitari a, 0451 O México, D.F. Teléfono y fax 5622-901 O. email: nfloresh @serviclo r.un am.mx y ava li @serv iclor.unam.mx 2 UBIPRO, ENEP-lztacala, Universidad Nacional Autónoma ele México. Apartado Postal 314, México, 54090, Tlaln epantla, Edo de México. 3 lnstituto de Biología, Universid ad Nacional Autónoma de México, Apartado Postal 70-233, Ciudad Universitaria, 04510 México, D.F. 'A utor para correspondenciaResumen. Este trabajo presenta una descripción de la vegetación esclerófi la perennifolia del Vall e de Tehuacán que es usada como base para compararla con 4 de las 5 zonas mediterrfü1eas. La descripc ión ele la vegetación se hizo considerando aspectos est ru ctura les tales como la cobertura, a ltura, frecuenc ia y espectros ele formas de vida. Esta vegetación denominada Mexical presenta es trechas similitudes con la vegetac ión ele climas mediterráneos en cuanto a los espectros de formas de vida y una predominancia de anua les, lo que complementa la hipótes is de que la distribución de la vegetación esc lerófi la perennifolia no só lo es resultado d e procesos asociados con e l clima mediterráneo, siendo exp li cab le su prese ncia en latitud es intertropical es principa lmente por factores históricos. Palabras clave: esclerófilo, perennifolio, conve rgencia, Va ll e de Te huacán, vegetación, Pu eb la. Abstract. The evergreen sclerophyllous vegetation of the Tehuacán Valley is desc ribed and co mparecl with 4 of the 5 zones with Mecliterranean-type clima tes. In the vegetation desc ription structura l aspects such as cove r, he ight, frequency ancl li fe forms spectrum were considered. This vegetation type ca lled Mexical has a close sim ilarity in life forms spectrum with annuals predominance with the Medite rranean-type vege tation. Therefore, the eve rgreen sclerophyllous vegetation distribution within the tropics not on ly is ex plain ecl in re lati o n with Mediterranean-type climates, ancl also because by historical factors .E 1 matorral esclerófilo perennifolio está constituido principalmente por arbustos achaparrados de 1 a 2 m de a lto con hojas esclerófilas perennifolias, además de coberturas cercanas al 100 % (Rzedowski, 1978). Los principales elementos que componen esta vegetación son arbustos de los géneros Arctostaphylos, Quercus, Rhu.s, Cercocmpu.s, Cotoneaster, etc. (Miranda y Hernández X, 1963). En México, este matorral tiene su principal representante en Baja California, e n la Sierra de Juárez y San Pedro Mártir, en un clima de tipo mediterráneo con lluvias en invierno y veranos secos; ocupa también una amplia extensión en California donde se le denomina Chaparral. La

Manuscript symbols can be stored, recognized and retrieved from an entropic digital memory that is associative and distributed but yet declarative; memory retrieval is a constructive operation; symbols not contained in the memory are rejected directly without search; and memory operations can be performed through parallel computations. Manuscript symbols, both letters and numerals, are stored in Associative Memory Registers that have an associated entropy. The memory recognition operation obeys an entropy trade-off between precision and recall, and the entropy level impacts on the quality of the objects recovered through the memory retrieval operation. We discuss the operational characteristics of the system for retrieving objects with both complete and incomplete information, such as severe occlusions. The experiments reported in this paper add evidence that supports the scalability of the framework and its potential for developing practical applications. We also compare the present entropic associative memories to Hopfield’s paradigm and discuss its potential for the study of natural memory.