Much of the ongoing explosion of digital content is in the form of text. This content is a virtual gold-mine of information that can inform a range of social, governmental, and business decisions. For example, using content available on blogs and social networking sites businesses can find out what its customers are saying about their products and services. In the digital age where customer is king, the business value of ascertaining consumer sentiment cannot be overstated. People express sentiments in myriad ways. At times, they use simple, direct assertions, but most often they use sentences involving comparisons, conjunctions expressing multiple and possibly opposing sentiments about multiple features and entities,and pronominal references whose resolution requires discourse level context. Frequently people use abbreviations, slang, SMSese, idioms and metaphors. Understanding the latter also requires common sense reasoning. In this paper, we present iSEE, a fully implemented sentiment extraction engine, which makes use of statistical methods, classical NLU techniques, common sense reasoning, and probabilistic inference to extract entity and feature specific sentiment from complex sentences and dialog. Most of the components of iSEE are domain independent and the system can be generalized to new domains by simply adding domain relevant lexicons.

The SHRUTI model demonstrates how a neurally motivated connectionist architecture can support encoding of complex evidential and relational knowledge and perform rapid inference and decision-making with that knowledge. Key to the successful operation of the SHRUTI system is its encoding of causal rules and its capacity to handle multiple instantiation of relational predicates. We present here a detailed description of the implementation of causal rules and of the manner in which multiple instantiation is achieved within the current SHRUTI architecture. The implementation described here represents an improvement over earlier realizations of similar functionality. In particular, it solves the multiple instantiation problem using a significantly simpler connectionist circuit.

All facets of high-level cognition, including perception, language, memory, and action, involve relational information (e.g., What is where? and Who did what to whom?). In order to process relational information, the brain must be capable of supporting a dynamic, activation-based expression of such information as well as a persistent, structure-based encoding of such information. An important challenge for computational neuroscience is to model the brain's remarkable ability to express and seamlessly integrate both transient and persistent forms of relational information. A brief review of work done by the author and his collaborators on computational models that attempt to address this challenge using temporal synchrony, network structure, and synaptic plasticity is presented.

An automatic transcription system has been developed to label and segment phonetic constituents of spontaneous American English without benefit of a word-level transcript. Instead, special-purpose neural networks classify each 10-ms frame of speech in terms of articulatory-acoustic-based phonetic features and the feature clusters are subsequently mapped to phoneticsegment labels using multilayer perceptron networks. The phonetic labels generated by this system are 80 % concordant with the labels produced by human transcribers and the segmental boundaries deviate from manual segmentation by an average of 11 ms. The automatic transcription system thus generates phonetic labels and segmentation comparable in quality to those produced by human transcribers, and therefore may prove useful for phonetic annotation of novel linguistic corpora, as well as facilitating development of pronunciation models for automatic speech recognition systems. 1.

Incremental class learning (ICL) presents a possible solution to the catastrophic interference problem and provides a framework for the development of scalable learning systems. With respect to multi-class classification problems, the ICL approach can be summarized as follows. Initially the system focuses on one category. After it learns this category, it tries to identify a compact subset of features (nodes) in the hidden layers, that are crucial for the recognition of this category. The system then freezes these crucial nodes (features) by fixing their incoming weights. As a result, these features cannot be obliterated in subsequent learning. Moreover, these frozen features are available during subsequent learning and can be shared among a number of categories. Finally, as more categories are learned, the set of features gradually stabilizes and learning a new category requires less effort. We present results of applying the ICL approach to the handwritten digit recognition problem.

The neural network learning algorithm presented in the paper splits the problem of handwritten digit recognition into easy steps by learning character classes incrementally: at each step, the neurons most relevant to the considered class are fixed so that subsequent classes will not disrupt the knowledge already acquired, but will be able to use it. A new relevance measure is also defined, for which a cheap approximation can be computed. The advantage of the attribution scheme starts to show even for small experiments, but should become more obvious as the number of classes increases. Picking only a few relevant features for each class, and sharing them between classes, constrains learning and improves generalization. Experiments were limited to pre-segmented digits, but our use of a spatio-temporal network architecture makes their extension to unsegmented strings straightforward.

We describe an alternate approach to visual recognition of handwritten words, wherein an image is converted into a spatio-temporal signal by scanning it in one or more directions, and processed by a suitable connectionist network. The scheme offers several attractive features including shift-invariance, explication of local spatial geometry along the scan direction, a significant reduction in the number of free parameters, the ability to process arbitrarily long images along the scan direction, and a natural framework for dealing with the segmentation/recognition dilemma. Other salient features of the work include the use of a modular and structured approach for network construction and the integration of connectionist components with a procedural component to exploit the complementary strengths of both techniques. The system consists of two connectionist components and a procedural controller. One network concurrently makes recognition and segmentation hypotheses, and another performs refined recognition of segmented characters. The interaction between the networks is governed by the procedural controller. The system is tested on three tasks: isolated digit recognition, recognition of overlapping pairs of digits and recognition of ZIP codes.

A connectionist model for recognizing unconstrained handprinted digits is described. Instead of treating the input as a static signal, the image is canned over time and converted into a time-varying signal. The temporalized image is processed by a spatiotemporal connectionist network. The resulting system offers shift-invariance along the temporalized axis, a reduction in the number of free parameters, and the ability to process images of arbitrary length. For a set of real-world ZIP code digit images, the system achieved a 99.1% recognition rate on the training set and a 96.0% recognition rate on the test with no rejections. A 99.0% recognition rate on the test set was achieved when 14.6% of the images were rejected.>

A fundamental problem that must be addressed by connectionism is that of creating and representing dynamic structures (Feldman 1982; von der Malsburg 1985). In the context of reasoning with systematic and abstract knowledge, this problem takes the form of the variable binding problem. We describe a biologically plausible solution to this problem and outline how a knowledge representation and reasoning system can use this solution to perform a class of predictive inferences with extreme efficiency. The proposed system solves the variable binding problem by propagating rhythmic patterns of activity wherein dynamic bindings are represented as the synchronous firing of appropriate nodes.

Many knowledge-based systems express domain knowledge in terms of a hierarchy of concepts (or frames)—where each concept is a collection of property value (or slot filler) pairs. One can associate two interesting classes of inference with such information structures: inheritance and recognition. Attempts at formalizing inheritance and recognition, however, have been confounded by the presence of conflicting property values among related concepts. Such conflicting information gives rise to the problems of exceptions and multiple inheritance during inheritance, and partial matching during recognition. This paper presents a formalization of inheritance and recognition based on the principle of maximum entropy. The proposed formalization offers several advantages: it admits necessary as well as default property values, it deals with conflicting information in a principled manner, and it solves the problems of exceptions, multiple inheritance, as well as partial matching. It can also be shown that the proposed formalization may be realized as a massively parallel network of simple processing elements that can solve an interesting class of inheritance and recognition problems extremely fast—in time proportional to the depth of the conceptual hierarchy.