Abstract: Methods for recognizing a category of an object are disclosed. In one embodiment, a method includes determining, by a processor, a preliminary category of a target object, the preliminary category having a confidence score associated therewith, and comparing the confidence score to a learning threshold. If the highest confidence score is less than the learning threshold, the method further includes estimating properties of the target object and generating a property score for one or more estimated properties, and searching a supplemental image collection for supplemental image data using the preliminary category and the one or more estimated properties. Robots programmed to recognize a category of an object by use of supplemental image data are also disclosed.

Abstract: Analysis of the audio component in multimedia data is disclosed. Rooms can be described through room impulse responses (RIR), the “fingerprint” of a specific room. The method uses machine learning techniques to identify rooms from ordinary audio recordings.

Abstract: The present invention successfully combines neural-net discriminative feature processing with Gaussian-mixture distribution modeling (GMM). By training one or more neural networks to generate subword probability posteriors, then using transformations of these estimates as the base features for a conventionally-trained Gaussian-mixture based system, substantial error rate reductions may be achieved. The present invention effectively has two acoustic models in tandem—first a neural net and then a GMM. By using a variety of combination schemes available for connectionist models, various systems based upon multiple features streams can be constructed with even greater error rate reductions.

Abstract: Methods and apparatus, including systems and computer program products, for implementing environments that provide user groups with the ability to post, discuss, and manage group content, and the ability to organize and make group decisions, based on the content of documents. One method includes managing documents according to a respective document state for each document in the environment. The possible document states include a submitted state, a voting state, an approved state, and a not approved state. A group member submitting a document to the group places the document in the submitted state. When the group coordinator enables voting on the document, it moves to the voting state. If the members approve the document, it moves from the voting to the approved state; and otherwise, the document moves from the voting to the not approved state.

Abstract: Systems and methods for executing electronic transactions on an anonymous basis using blind auditable membership proofs. By making use of a new cryptographic primitive, electronic transactions such as payment, voting, investment, redemption of tax coupon, and international currency transfer may be made both anonymous and auditable. In an electronic payment system according to the present invention, a user submits information identifying a coin to a bank which in turn validates the coin and adds it to a public list of valid coins. To make a payment using the coin, the user presents an efficient auditable membership proof to a merchant in the form of a zero knowledge argument which proves that the user knows the authenticating information for an unspecified coin in the public list of valid coins. The merchant verifies the zero knowledge argument, accepts the coin as payment, and presents certain authenticating information to the bank.

Abstract: An automatic speech recognition apparatus and method with a front end feature extractor that improves recognition performance under adverse acoustic conditions are disclosed. The inventive feature extractor is characterized by a critical bandwidth spectral resolution, an emphasis on slow changes in the spectral structure of the speech signal, and adaptive automatic gain control. In one the feature extractor includes a feature generator configured to compute short-term parameters of the speech signal, a filter system configured to filter the time sequences of the short-term parameters, and a normalizer configured to normalize the filtered parameters with respect to one or more previous values of the filtered parameters.

Abstract: An online conference session management system includes computer sites associated with a conference session holder, a plurality of service providers and a plurality of conference session participants. The conference session holder and each service provider has a respective associated public/private key pair for a specified online conference session. Each service provider is provided with an active participant registration certificate that gives the service provider the authority to register participants of the specified conference session, as well as authority to register other service providers as active participants of the specified conference session by providing those other service providers with respective active participant registration certificates.

Abstract: A data processing system contains both a scalar processor and a vector processor. The vector processor contains a plurality of functional units, each of which contains a plurality of parallel pipelines. Each of the pipelines contains a plurality of arithmetic and logic units (ALUs) connected via a plurality of data paths, such that data can be communicated between the ALUs during the execution of a vector instruction by the vector functional unit containing the pipeline. The operation performed by each of the cascaded ALUs and the paths through which data is to be communicated between the ALUs during the execution of a vector instruction can be controlled by configuration values held in a scalar register named by the vector instruction. Through the use of this technique, multiple operations upon sets of vector data may be specified in a single short vector instruction, and further, the configuration of the pipelines can be determined dynamically in response to program input.

Abstract: A data distribution system has a number of data processing devices interconnected by data transmission media. At least one of the data processing devices transmit data to other data processing devices in the system. The transmitting data processing system includes memory for storing the data to be transmitted, programmable data processing circuitry, and data transmission apparatus for transmitting an encoded representation of the stored message as a sequence of data packets. Priority data, stored in the memory, represents a plurality of assigned priority levels for specified portions of the stored message such that portions of the stored message have respective assigned priority levels. A data encoding program generates an encoded representation of the stored message that includes, for each portion of the stored message, a level of redundant data corresponding to the priority level assigned to that portion of the stored message.

Abstract: A method and system are provided for alleviating the harmful effects of convolutional and additive noise in speech, such as due to the environmental noise and linear spectral modification, based on the filtering of time trajectories of an auditory-like spectrum in a particular spectral domain.

Abstract: A method and system are provided for alleviating the harmful effects of convolutional distortions of speech, such as the effect of a telecommunication channel, on the performance of an automatic speech recognizer (ASR). The technique is based on the filtering of time trajectories of an auditory-like spectrum derived from the Perceptual Linear Predictive (PLP) method of speech parameter estimation.