We present a hybrid neural-network for human face recognition which compares favourably with other methods. The system combines local image sampling, a self-organizing map (SOM) neural network, and a convolutional neural network. The SOM provides a quantization of the image samples into a topological space where inputs that are nearby in the original space are also nearby in the output space, thereby providing dimensionality reduction and invariance to minor changes in the image sample, and the convolutional neural network provides partial invariance to translation, rotation, scale, and deformation. The convolutional network extracts successively larger features in a hierarchical set of layers. We present results using the Karhunen-Loeve transform in place of the SOM, and a multilayer perceptron (MLP) in place of the convolutional network for comparison. We use a database of 400 images of 40 individuals which contains quite a high degree of variability in expression, pose, and facial details. We analyze the computational complexity and discuss how new classes could be added to the trained recognizer.

The vast improvement in information access is not the only advantage resulting from the increasing percentage of hyperlinked human knowledge available on the Web. Additionally, much potential exists for analyzing interests and relationships within science and society. However, the Web's decentralized and unorganized nature hampers content analysis. Millions of individuals operating independently and having a variety of backgrounds, knowledge, goals and cultures author the information on the Web. Despite the Web's decentralized, unorganized, and heterogeneous nature, our work shows that the Web self-organizes and its link structure allows efficient identification of communities. This self-organization is significant because no central authority or process governs the formation and structure of hyperlinks.

It has previously been shown that gradient-descent learning algorithms for recurrent neural networks can perform poorly on tasks that involve long-term dependencies, i.e. those problems for which the desired output depends on inputs presented at times far in the past. We show that the long-term dependencies problem is lessened for a class of architectures called nonlinear autoregressive models with exogenous (NARX) recurrent neural networks, which have powerful representational capabilities. We have previously reported that gradient descent learning can be more effective in NARX networks than in recurrent neural network architectures that have "hidden states" on problems including grammatical inference and nonlinear system identification. Typically, the network converges much faster and generalizes better than other networks. The results in this paper are consistent with this phenomenon. We present some experimental results which show that NARX networks can often retain information for two to three times as long as conventional recurrent neural networks. We show that although NARX networks do not circumvent the problem of long-term dependencies, they can greatly improve performance on long-term dependency problems. We also describe in detail some of the assumptions regarding what it means to latch information robustly and suggest possible ways to loosen these assumptions.

The coverage and recency of the major World Wide Web search engines was analyzed, yielding some surprising results. The coverage of any one engine is significantly limited: No single engine indexes more than about one-third of the “indexable Web,” the coverage of the six engines investigated varies by an order of magnitude, and combining the results of the six engines yields about 3.5 times as many documents on average as compared with the results from only one engine. Analysis of the overlap between pairs of engines gives an estimated lower bound on the size of the indexable Web of 320 million pages.