Abstract: Described is a technology by which uploaded GPS data is indexed according to spatio-temporal relationships to facilitate efficient insertion and retrieval. The indexes may be converted to significantly smaller-sized data structures when new updates to that structure are not likely. GPS data is processed into a track of spatially-partitioned segments such that each segment has a cell. Each cell has an associated temporal index (a compressed start-end tree), into which data for that cell's segments are inserted. The temporal index may include an end time index that relates each segment's end time to a matching start time index. Given query input comprising a spatial predicate and a temporal predicate, tracks may be searched for by determining which spatial candidate cells may contain matching results. For each candidate cell, the search accesses the cell's associated temporal index to find any track or tracks that correspond to the temporal predicate.

Abstract: The simultaneous localization and RF modeling technique pertains to a method of providing simultaneous localization and radio frequency (RF) modeling. In one embodiment, the technique operates in a space with wireless local area network coverage (or other RF transmitters). Users carrying Wi-Fi-enabled devices traverse this space while the mobile devices record the Received Signal Strength (RSS) measurements corresponding to access points (APs) in view at various unknown locations and report these RSS measurements, as well as nay other available location fix to a localization server. A RF modeling algorithm runs on the server and is used to estimate the location of the APs using the recorded RSSI measurements and any other available location information. All of the observations are constrained by the physics of wireless propagation. The technique models these constraints and uses a genetic algorithm to solve them, thereby providing an absolute location of the mobile device.

Abstract: A packet forwarding technique is presented that forwards data packets through a multi-hop wireless network employing multi-radio nodes. In each multi-radio node, an output buffer is shared among the radios. Outgoing packets are stored in this buffer. Whenever there is a transmission opportunity on the radios sharing the output buffer, the node examines the buffer and searches for packets that are scheduled to be transmitted to a neighboring node having an open communication link with the transmitting node. The packet in the discovered group that has the highest transmission priority is then transmitted to its next scheduled node.

Abstract: After an initial training session, a “Dynamic Echo Canceller” (DEC) provides echo cancellation where only access to an input signal and a composite output signal are available, and the input signal is subjected to an unknown variable gain function. In one embodiment, the DEC uses echo cancellation to provide a “clean” copy of a second input signal where only a first input signal and a composite of the first and second input signal is available. An example is a “black box” amplifier coupled to a microphone and a phone line, with access to only the microphone input and a combined output signal where it is desired to retrieve a clean copy of a remote caller signal from the combined output. The DEC is applicable to many fields, including: signal separation; cancellation of echoes caused by impedance mismatches, periodic electrical noise, acoustic echoes caused by acoustic coupling, etc.

Abstract: Audio in an audio conference is spatialized using either virtual sound-source positioning or sound-field capture. A spatial audio conference is provided between a local and remote parties using audio conferencing devices (ACDs) interconnected by a network. Each ACD captures spatial audio information from the local party, generates either one, or three or more, audio data streams which include the captured information, and transmits the generated stream(s) to each remote party. Each ACD also receives the generated audio data stream(s) transmitted from each of the remote parties, processes the received streams to generate a plurality of audio signals, and renders the signals to produce a sound-field that is perceived by the local party, where the sound-field includes the spatial audio information captured from the remote parties. A sound-field capture device is also provided which includes at least three directional microphones symmetrically configured about a center axis in a semicircular array.

Abstract: The page ranking technique described herein employs a Markov Skeleton Mirror Process (MSMP), which is a particular case of Markov Skeleton Processes, to model and calculate page importance scores. Given a web graph and its metadata, the technique builds an MSMP model on the web graph. It first estimates the stationary distribution of a EMC and views it as transition probability. It next computes the mean staying time using the metadata. Finally, it calculates the product of transition probability and mean staying time, which is actually the stationary distribution of MSMP. This is regarded as page importance.

Abstract: A method and system for synthesizing texture using upsampled pixel coordinates and a multi-resolution approach. The parallel texture synthesis technique, while based on a neighborhood matching technique having order-independent texture synthesis, extends that approach in at least two areas, including efficient parallel synthesis and intuitive user control. Pixel coordinates are upsampled instead of pixel colors, thereby reducing computational complexity and expense. These upsampled pixel coordinates then are jittered to provide texture variation. The jitter is controllable, such that a user has control over several aspects of the jitter. In addition, each neighborhood-matching pass is split into several sub-passes to improve correction. Using sub-passes improves correction speed and quality. The parallel texture synthesis system and method disclosed herein is designed for implementation on a parallel processor, such as a graphics processing unit.

Abstract: A word alignment modeler uses probabilistic learning techniques to train “word-dependent transition models” for use in constructing phrase level Hidden Markov Model (HMM) based word alignment models. As defined herein, “word-dependent transition models” provide a probabilistic model wherein for each source word in training data, a self-transition probability is modeled in combination with a probability of jumping from that particular word to a different word, thereby providing a full transition model for each word in a source phrase. HMM based word alignment models are then used for various word alignment and machine translation tasks. In additional embodiments sparse data problems (i.e., rarely used words) are addressed by using probabilistic learning techniques to estimate word-dependent transition model parameters by maximum a posteriori (MAP) training.

Abstract: A full-text lattice indexing and searching system and method for indexing word lattices using a text indexer to enable enhance searching of audio content. The system and method utilize a Time-Anchored Lattice Expansion (TALE) method that represents word lattices such that they can be indexed with existing text indexers with little or no modification. Embodiments of system and method include an indexing module for generating and indexing word lattices based on audio content and a searching module for allowing searching of a full-text index containing indexed word lattices. The indexing module includes a custom IFilter and a custom Wordbreaker. Embodiments of the searching module include an ExpandQuery function for decorating an input query and a custom Stemmer. Embodiments of the searching module also include a GenerateSnippets module that extracts information from the indexed word lattices to enable the creation of clickable snippets.

Abstract: A streaming operator assignment system and method for determining a streaming operator assignment that minimizes overload in a data processing system. Embodiments of the streaming operator assignment system include an optimization goals definition module, which defines optimization goals in terms of fundamental quantities that system administrators and application writers want to control, such as minimizing the worst case latency over all periods of time, or minimizing how much the system is backlogged with work. Embodiments of the streaming operator assignment system also include an optimization goals solution module that optimizes and solves a selected optimization goal. A specialized optimization technique is used to find the best operator (or load) assignment using the optimization goals to measure of the value of the assignment.

Abstract: An erasure resilient coding (ERC) distributed data storage system and method for storing data in a reliable and survivable fashion while minimizing hardware and associated costs. The system and method includes forming multiple protection groups both within and across storage nodes of the storage system. Data is segmented into original data blocks and ERC data blocks. Load balancing occurs by interleaving storage nodes with equal numbers of original data blocks and ERC data blocks while ensuring each node has an equal number of combined read and write operations. Unique read and write operations on data block can be performed independent of other data blocks in a protection group. The write operation uses Galois field arithmetic and ERC transform to either write or append a new data block to a storage node. The read operation recovers data in a variety of ways using ERC decoding.

Abstract: A local network coding framework and method including techniques to improve efficiency in a wireless network by reducing overhead. The local network coding method includes exchanging data availability between nodes on the wireless network by sending Bloom filters of lists of packets to neighboring nodes. Based on data availability, optimized mixing of pure packets is performed to form mixture packets for output. A separate acknowledgement buffer keeps track of the pure packets transmitted but not acknowledged. If an acknowledgement does not arrive after a certain time period, the packet is assumed to be lost and is retransmitted. An optimized packet mixing process generates mixture packets and decides which nodes to send the mixture packets. The local network coding framework and method also includes methods for representing the composition of a mixture packet and using mixing at a wireless access point to improve the performance of the wireless local area network.

Abstract: A video noise reduction technique is presented. Generally, the technique involves first decomposing each frame of the video into low-pass and high-pass frequency components. Then, for each frame of the video after the first frame, an estimate of a noise variance in the high pass component is obtained. The noise in the high pass component of each pixel of each frame is reduced using the noise variance estimate obtained for the frame under consideration, whenever there has been no substantial motion exhibited by the pixel since the last previous frame. Evidence of motion is determined by analyzing the high and low pass components.

Abstract: A transformation-based record matching technique. The technique provides a flexible way to account for synonyms and more general forms of string equivalences when performing record matching by taking as explicit input user-defined transformation rules (such as, for example, the fact that “Robert” and “Bob” that are synonymous). The input string and user-defined transformation rules are used to generate a larger set of strings which are used when performing record matching. Both the input string and data elements in a database can be transformed using the user-defined transformation rules in order to generate a larger set of potential record matches. These potential record matches can then be subjected to a threshold test in order to determine one or more best matches. Additionally, signature-based similarity functions are used to improve the computational efficiency of the technique.

Abstract: A “Classifier Trainer” trains a combination classifier for detecting specific objects in signals (e.g., faces in images, words in speech, patterns in signals, etc.). In one embodiment “multiple instance pruning” (MIP) is introduced for training weak classifiers or “features” of the combination classifier. Specifically, a trained combination classifier and associated final threshold for setting false positive/negative operating points are combined with learned intermediate rejection thresholds to construct the combination classifier. Rejection thresholds are learned using a pruning process which ensures that objects detected by the original combination classifier are also detected by the combination classifier, thereby guaranteeing the same detection rate on the training set after pruning. The only parameter required throughout training is a target detection rate for the final cascade system.

Abstract: A subregion-based image parameter recovery system and method for recovering image parameters from a single image containing a face taken under sub-optimal illumination conditions. The recovered image parameters (including albedo, illumination, and face geometry) can be used to generate face images under a new lighting environment. The method includes dividing the face in the image into numerous smaller regions, generating an albedo morphable model for each region, and using a Markov Random Fields (MRF)-based framework to model the spatial dependence between neighboring regions. Different types of regions are defined, including saturated, shadow, regular, and occluded regions. Each pixel in the image is classified and assigned to a region based on intensity, and then weighted based on its classification.

Abstract: A “Client-Side Echo Canceller” provides a unique system and method for reducing Multipoint Control Unit (MCU) computational overhead in a multi-point audio conference. In general, the local audio input signal of each client is transmitted in real-time to the MCU. The MCU then combines the audio input signals of all clients to create a single composite signal that is transmitted back to all clients in real-time. Each client then locally processes the composite signal to remove each client's local contribution to the composite signal prior to local playback in order to eliminate a local echo of each client's local audio input. In various embodiments, local cancellation of the local audio input from the composite signal is performed on either a time domain or a transform domain representation of the composite signal. Further, since each client receives the same signal, MCU transmission bandwidth can be reduced via multicast transmissions.

Abstract: A novel adaptive beamforming technique with enhanced noise suppression capability. The technique incorporates the sound-source presence probability into an adaptive blocking matrix. In one embodiment the sound-source presence probability is estimated based on the instantaneous direction of arrival of the input signals and voice activity detection. The technique guarantees robustness to steering vector errors without imposing ad hoc constraints on the adaptive filter coefficients. It can provide good suppression performance for both directional interference signals as well as isotropic ambient noise.

Abstract: A novel beamforming post-processor technique with enhanced noise suppression capability. The present beam forming post-processor technique is a non-linear post-processing technique for sensor arrays (e.g., microphone arrays) which improves the directivity and signal separation capabilities. The technique works in so-called instantaneous direction of arrival space, estimates the probability for sound coming from a given incident angle or look-up direction and applies a time-varying, gain based, spatio-temporal filter for suppressing sounds coming from directions other than the sound source direction resulting in minimal artifacts and musical noise.

Abstract: A search expertise level system and method for determining a search expertise level of a search engine user and then using that information to improve the searcher's experience. The search expertise level system and method identifies the search expertise level of the searcher based on query behavior, post-query browsing behavior, and other behaviors of the searcher. One simple and important behavior that indicates a skilled searcher is the use of advanced query syntax and operators in the query. Once the search expertise level of a searcher is known, the search engine user interface can be modified and tailored to the needs of both skilled and novice searchers. The search expertise level also can be used to rank search results, such that search results for a novice searcher are ranked differently than those for a skilled searcher. The search expertise level also can be used in advertising and marketing.