Book reviews

Traditionelle und moderne Formen der Landwirtschaft in Iran: Siedlung, Wirtschaft und Agrarsozialstruktur in nördlichen Khuzistan seit dem Ende des 19. Jahrhunderts. [Traditional and Modern Forms of Agriculture in Iran: Settlement, Economics, and Agrarian Social Structure in Northern Khuzistan since the End of the Nineteenth Century.] By Eckhart Ehlers. With a contribution by Grace Goodell, “Agricultural Production in a Traditional Village of Northern Khuzistan.” A publication of the Geographical Institute of Marburg University in the series, Marburger geographische Schriften, Vol. 64, 1975. x + 289 pp., 12 charts, DM33.

Unity in the Ghazals of Hafez. By Michael C. Hillmann. Minneapolis and Chicago: Bibliotheca Islamica, 1976. 181 pp.

Afghanistan in the 1970s. Edited by Louis Dupree and Linette Albert. New York: Praeger Publishers, 1974.     

Book reviews

The Agguyunlu: Clan, Confederation, Empire. A Study in 15th/9th Century Turko‐Iranian Politics. By John E. Woods. (Studies in Middle Eastern History, No. 3), xv + 348 pp. Minneapolis and Chicago: Bibliotheca Islamica, 1976.

A Persian Stronghold of Zoroastrianism. By Mary Boyce. New York: Oxford University Press, 1977. 284 pp. $19.95.

The Crowned Cannibals: Writings on Repression in Iran. By Reza Baraheni. With an introduction by E. L. Doctorow. New York: Random House/Vintage Books, 1977. 279 pp. $3.95.

Tarikh: Bulletin of the Department of History, Faculty of Letters and Humanities. Tehran: University of Tehran. Vol. 1, Nos. 1–2 (1976–77).

Middle East Economies in the 1970's: A Comparative Approach. By Hossein Askari and John Thomas Cummings. New York: Praeger Publishers (Praeger Special Studies in International Economics and Development). 1976. xxxiii + 581pp.

The Caspian Circle: A Novel. By Donné Raffat. Boston: Houghton Mifflin, 1978. 350 pp. $10.

The Agricultural Development of Iran. By Oddvar Aresvik. New York: Praeger, 1976. 271 pp.

Dschalaluddin Rumi, Licht und Reigen: Gedichte aus dem Divan des grössten mystischen Dichters persischer Zunge. Selected, translated and explained by Johann Cristoph Burgel. UNESCO Collection of Representative Works. Masterpieces of Persian Literature Series, No. 26. Bern, Herbert Lang, and Frankfurt/Main: Peter Lang, 1974. 191 pp.     

Book reviews

Modern Iran: The Dialectics of Continuity and Change. Edited by Michael E. Bonine and Nikki R. Keddie. Albany: State University of New York Press, 1981. x, 464 pp. $34.00.

IRANIAN SOCIETY. An Anthology of Writings by Jalal Al‐e Ahmad. Compiled and Edited by Michael C. Hillmann. Lexington, Kentucky: Mazda Publishers, 1982. 157 pp.

Modern Persian Short Stories. Trans. Minoo S. Southgate. Washington, D.C.: Three Continents Press, 1980. xvi + 228 pp.

Moderne Erzähler der Welt—Iran (Modern Storytellers of the World—Iran). Selected and edited by Faramarz Behzad, Johann Christoph Burgel, Gottfried Herrmann. Volume LVII Series on Cultural Encounters of the Institute for Foreign Relations, Stuttgart. Tubingen and Basel: Horst Erdmann Publishers for International Cultural Exchange, 1978.

La poésie persane après la seconde guerre mondiale. By Franciszek Machalski. Krakow: Polish Academy of Sciences, 1980. 133 pp.

Ja‐ye Khali‐ye Soluch. By Mahmud Dowlatabadi. Tehran: Agah, 1979. 497 pp.

History of Persian Literature from the Beginning of the Islamic Period to the Present Day. With contributions by George Morrison, Julian Baldick, and (Mohammad Reza) Sha‐fi'i Kadkani. Edited by George Morrison. Leiden‐Koln: E. J. Brill, 1981. 226 pp.     

Housing Choice in a Regulated Market: A Nested Multinomial Logit Analysis

The paper integrates several important elements of the institutionalized and regulated nature of the housing market and analyzes the relationship between household type and housing choice in Vienna within a nested multinomial logit framework. In particular, the concept of household specific choice sets is used to account for institutional, informational and income-based constraints of the choice process. Government subsidies (such as housing and rent allowances, non-interest-bearing state loans) are explicitly taken into account in the definition of variables. Housing choice is considered in three stages: the choice of a dwelling unit given dwelling type and residential zone, the choice of a dwelling type given residential zone, and the marginal choice of a residential zone. The coefficients are derived by means of a sequential ML-procedure. The empirical results clearly indicate that demographic variables have significant impacts on housing choice behavior. Income as a single explanatory variable as well as its interacting with size measures and the quality of the house are found to be important criteria for dwelling unit choice, as housing cost variables for dwelling type choice behavior. The dwelling unit and dwelling type choice submodels fit reasonably well, whereas the residential zone model is less successful.     

Optimization in an Error Backpropagation Neural Network Environment with a Performance Test on a Spectral Pattern Classification Problem

This paper attempts to develop a mathematically rigid framework for minimizing the cross-entropy function in an error backpropagating framework. In doing so, we derive the backpropagation formulae for evaluating the partial derivatives in a computationally efficient way. Various techniques of optimizing the multiple-class cross-entropy error function to train single hidden layer neural network classifiers with softmax output transfer functions are investigated on a real-world multispectral pixel-by-pixel classification problem that is of fundamental importance in remote sensing. These techniques include epoch-based and batch versions of backpropagation of gradient descent, PR-conjugate gradient, and BFGS quasi-Newton errors. The method of choice depends upon the nature of the learning task and whether one wants to optimize learning for speed or classification performance. It was found that, comparatively considered, gradient descent error backpropagation provided the best and most stable out-of-sample performance results across batch and epoch-based modes of operation. If the goal is to maximize learning speed and a sacrifice in classification accuracy is acceptable, then PR-conjugate gradient error backpropagation tends to be superior. If the training set is very large, stochastic epoch-based versions of local optimizers should be chosen utilizing a larger rather than a smaller epoch size to avoid unacceptable instabilities in the classification results.