Advanced Network Model Research Articles

Use of Artificial Neural Network in Differentiation of Subgroups of Temporomandibular Internal Derangements: A Preliminary Study

Artificial neural networks (ANNs) have been developed in the past few decades for many different applications in medical science and in biomedical research. The use of neural networks in oral and maxillofacial surgery is limited. The aim of this study was to determine the use of ANNs for the prediction of 2 subgroups of temporomandibular joint (TMJ) internal derangements (IDs) and normal joints using characteristic clinical signs and symptoms of the diseases. Clinical symptoms and diagnoses of 161 patients with TMJ ID were considered the gold standard and were employed to train a neural network. After the training process, the symptoms and diagnoses of 58 new patients were used to verify the network's ability to diagnose. The diagnoses obtained from ANNs were compared with diagnoses of a surgeon experienced in temporomandibular disorders. The sensitivity and specificity of ANNs in predicting subtypes of TMJ ID were evaluated using clinical diagnosis as the gold standard. Eight cases evaluated as bilaterally normal in clinical examination were evaluated as normal by ANN. In detecting unilateral anterior disc displacement with reduction (ADDwR; clicking), the sensitivity and specificity of ANN were 80% and 95%, respectively. In detecting unilateral anterior disc displacement without reduction (ADDwoR; locking), the sensitivity and specificity of ANN were 69% and 91%, respectively. In detecting bilateral ADDwoR, the sensitivity and specificity of ANN were 37% and 100%, respectively. In detecting bilateral ADDwR, the sensitivity and specificity of ANN were 100% and 89%, respectively. In detecting cases of ADDwR at 1 side and ADDwoR at the other side, the sensitivity and specificity of ANN were 44% and 93%, respectively. The application of ANNs for diagnosis of subtypes of TMJ IDs may be a useful supportive diagnostic method, especially for dental practitioners. Further research, including advanced network models that use clinical data and radiographic images, is recommended.

Queueing Model Improves IBM's Semiconductor Capacity and Lead-Time Management

At IBM's 300 mm semiconductor plant in East Fishkill, New York, we developed and implemented the Enterprise Production Planning and Optimization System (EPOS), an advanced queueing network model for capacity planning. EPOS extends earlier queueing network models by adding the ability to model product-specific batch service and batch arrivals and multichamber process equipment, and by embedding a linear program for allocating lots to the right queues when route choices are present. EPOS addresses both short-term tactical capacity planning and long-term strategic capital investment planning. It has become an integral part of predicting factory bottlenecks, prioritizing continuous-improvement efforts, planning capital equipment investments, and managing factory lead times. EPOS results have guided IBM's efforts to improve factory performance and reduce tens of millions of dollars of expense.

Book Review: Network Analysis and Ethnographic Problems: Process Models of a Turkish Nomad Clan, by Douglas R. White and Ulla C. Johansen (Oxford, UK and Lanham, MD: Lexington Books, 2004, Cloth / 2006, Paper, 544 pages)

Readers of Structure & Dynamics will appreciate Network Analysis and Ethnographic Problems by Douglas White and Ulla Johansen because it takes the very best in mathematical approaches to social structure and applies them to real data of significant scale over a period of many decades. The book represents a perfect combination of ethnographic method, using detailed professional observations over many decades (1956-1995) with analysis using highly advanced network modeling, graph theory, and, in the estimation of this reviewer, unexcelled computational expertise. The association of three highly unusual factors—years of genealogical information, years of observing and recording a people’s marriages and migrations, and an interest in constructing scientific explanatory statements about the social structures that emerge as these natural and cultural processes unfold—results in a unique book. In theoretical interest, it has something in common with two foundational works that set off years of anthropological wrangling about kinship, George Peter Murdock’s Social Structure (1949) and Claude Levi-Strauss’ Elementary Structures of Kinship (1969, original French version in 1949). What seems to have prompted mathematical anthropologist and network analyst Douglas White to team up with more traditional ethnographer Ulla Johansen was the possibility of using her incredibly detailed knowledge of Turkish nomads’ social activities over several generations to try to resolve the most vexing problems of those two opposing theoretical systems. The result of the collaboration between White and Johansen is a book that, like those two earlier efforts, will undoubtedly serve as a take-off point for many other scholars to follow. In addition to its contribution to our understanding of kinship theory in a quite new way, this book makes an outstanding contribution by reintroducing ethnographers to the network perspective. Fifty years ago, social anthropologists J. A. Barnes and Elizabeth Bott introduced network analysis into ethnography. Their initial steps made it possible for J. Clyde Mitchell to cut an exciting path that was soon followed by other British social anthropologists, revealing how valuable network analysis could be, especially in the study of urbanization of African populations. The end of the formal colonial era slowed social anthropological research in Africa in the decades following 1960. Also, changes in the interests of anthropologists in the English-speaking world, such as the reduction of interest in comparative social structures and cultures, coupled with rising interest in qualitative interpretive ethnography, left the field of social network analysis to be developed by sociologists and organizational scientists. Douglas White, however, is one of a small number of anthropologists who developed the mathematical expertise necessary to make a genuine contribution to network analysis of complex systems through combining graph theory, long-term fieldwork, and the electronic computation required to deal with masses of quantitative data. These are the skills he brought to this work, putting this book at the cutting edge of network studies, regardless of discipline. So, even those who are not interested in Turkish nomads, per se, may find this book of great value. Whether the reader is interested in kinship, in economics, in politics or history, this book might be considered must reading. However, it isn’t easy reading. Although the book is well-written, the subject matter is very complex and multidimensional. It contains many necessary tables and figures. It even gives “url” addresses to color-coded figures that are available only on the web, figures so complex that they cannot be understood without the color coding. Something that some readers may not understand is that although most graphic illustrations appear in only two dimensions, graph theory is designed to explain

Short-Term Traffic Flow Prediction Using Neuro-Genetic Algorithms

This paper presents a new short-term traffic flow prediction system based on an advanced Time Delay Neural Network (TDNN) model, the structure of which is synthesized using a Genetic Algorithm (GA). The model predicts flow and occupancy values at a given freeway section based on contributions from their recent temporal profile (over a few minutes) as well the spatial profile (including inputs from neighboring upstream and downstream sections). An in-depth investigation of the variables pertinent to traffic flow prediction was conducted examining the extent of the "look-back" in time interval, the extent of prediction in the future, the extent of spatial contribution, the resolution of the input data, and their effects on prediction accuracy. The model's performance is validated using both simulated and real traffic flow data obtained from the ATMS Testbed in Orange County, California. Both temporal and spatial effects were found to be essential for proper prediction. Results obtained indicate that the prediction errors vary inversely with the extent of the spatial contribution, and that the inclusion of three loop stations in both directions of the subject station is sufficient for practical purposes. Also, the longer the extent of prediction, the more the predicted values tend toward the mean of the actual, for which case the optimal look-back interval also shortens. The results also indicate that the level of data aggregation/resolution should be comparable to the prediction horizon for best accuracy. The model performed acceptably using both simulated and real data. The model also showed potential to be superior to such other well-known neural network models as the Multi layer Feed-forward (MLF) when applied to the same problem.

A hybrid paradigm for modeling of complex systems

The network modeling approach to simulation provides the mo deler with simple and powerful concepts which can be used to capture the significant aspects of the system to be modeled. Yet the most advanced network modeling environments lack ex plicit concepts for the representation of complex behavior such as decision-making. Artificial intelligence research, because of its emphasis on knowledge representation, provides several meth odologies which can be successfully applied to the modeling of decision-making behavior. The approach to modeling complex behavior, outlined in this paper, is based on a hybrid methodology unifying the concepts of Object-Oriented programming, Logic Programming and the discrete event approach to systems modeling. We present SIM YON, an experimental network simulation environment which provides explicit constructs for the representation of complex behavior of real-world systems. SIMYON is implemented by de fining a library of logic objects in the Object-Oriented, Logic Programming environment CAYENE. These objects, which are analogous to the "nodes" of network simulation languages, are the building blocks for modeling. Examples are given in SIMYON to model a job scheduler in a manufacturing situation and an adaptive material handling dis patch mechanism for flexible manufacturing systems.