Rumelhart D.E., McClelland J.L. Parallel Distributed Processing. Explorations in the Microstructure of Cognition. Volume 1. Foundations

Издательство MIT Press, 1999, -564 pp.One of the great joys of science lies in the moment of shared discovery. One person's half-baked suggestion resonates in the mind of another and suddenly takes on a definite shape. An insightful critique of one way of thinking about a problem leads to another, better understanding. An incomprehensible simulation result suddenly makes sense as two people try to understand it together.
This book grew out of many such moments. The seeds of the book were sown in our joint work on the interactive activation model of word perception. Since then, each of us has worked with the other and with other collaborators. The results of these collaborations are reported in several of the chapters of this book. The book also contains many chapters by other colleagues whose explorations have become intertwined with ours. Each chapter has its own by-line, but each also reflects the influences of other members of the group. We hope the result reflects some of the benefits of parallel distributed processing!
The idea of parallel distributed processing- the notion that intelligence emerges from the interactions of large numbers of simple processing units - has come and gone before. The idea began to seem more and more attractive to us as the contrast between our convictions about basic characteristics of human perception, memory, language, and thought and the accepted formal tools for capturing mental processes became more apparent. Symbol-processing machines, for all their Turing equivalence, had failed to provide useful frameworks for capturing the simple insights about the interactive nature of processing that had led to such models as the HEARSAY model of speech understanding. More generally, they had failed to provide a framework for representing knowledge in a way that allowed it to be accessed by content and effectively combined with other knowledge to produce useful automatic syntheses that would allow intelligence to be productive. And they made no contact with the real strengths and weaknesses of the hardware in the brain. A Cray computer can perform on the order of 100 million double-precision multiplications in a second, but it does not exhibit natural intelligence. How then are we to understand the capabilities of human thought, given the time constants and noisiness inherent in neural systems? It seemed obvious that to get any processing done in real time, the slow, noisy hardware in the brain would have to do massively parallel processing.Volume 1 Foundations (/file/1781043/)
Part I The PDP Perspective
The Appeal of Parallel Distributed Processing
A General Framework for Parallel Distributed Processing
Distributed Representations
PDP Models and General Issues in Cognitive Science
Part II Basic Mechanisms
Feature Discovery by Competitive Learning
Information Processing in Dynamical Systems: Foundations of Harmony Theory
Learning and Relearning in Boltzmann Machines
Learning Internal Representations by Error Propagation
Part III Formal Analyses
An Introduction to Linear Algebra in Parallel Distributed Processing
The Logic of Activation Functions
An Analysis of the Delta Rule and the Learning of Statistical Associations
Resource Requirements of Standard and Programmable Nets
P3: A Parallel Network Simulating SystemVolume 2 Psychological and Biological Models (/file/1781045/)
Part IV Psychological Processes
Schemata and Sequential Thought Processes in PDP Models
Interactive Processes in Speech Perception: The TRACE Model
The Programmable Blackboard Model of Reading
A Distributed Model of Human Learning and Memory
On Learning the Past Tenses of English Verbs
Mechanisms of Sentence Processing: Assigning Roles to Constituents
Part V Biological Mechanisms
Certain Assess of the Anatomy and Physiology of the Cerebral Cortex
Open Questions about Computation in Cerebral Cortex
Neural and Conceptual Interpretation of PDP Models
Biologically Plausible Models of Place Recognition and Goal Location
State-Dependent Factors Influencing Neural Plasticity: A Partial Account of the Critical Period
Amnesia and Distributed Memory
Part VI Conclusion
Reflections on Cognition and Parallel Distributed Processing