Abstract: In a computer system having a hierarchical memory, the problem of tolerating cache miss latency is solved by dynamically switching appropriately between two different code sequences, one optimized at compile-time, assuming a cache-hit, and the other optimized at compile-time, assuming a cache-miss.

Abstract: A computerized method of building balanced ranges of attribute values for multiple attributes of a database simultaneously without requiring presorting of the database. The balanced ranges are used to construct balanced range-based multidimensional bitmap indexes for processing complex multipredicate queries against the database. A dynamic partition expansion and contraction method can construct balanced ranges for bitmap indexing of tuples having high cardinality attributes and even in the presence of highly skewed data.

Abstract: An efficient method and system of communicating and accumulating global information for parallel data mining. Since each database is partitioned, a single node in a parallel environment will be unable to make decisions based on global information. A novel technique of progressive thresholding is devised to accumulates global information as a function of a transaction support threshold while minimizing transmission overhead. The transaction support threshold is varied as a function of the number of nodes contributing to the global information progresses.

Abstract: A memoryless communications adapter for communicating between the nodes of a distributed parallel computer network, each node including a non-shared program memory coupled to the memoryless communications adapter which interfaces the network. An embodiment of the present invention advantageously provides an adapter that can manipulate queues and matching tables efficiently in hardware and provide a high-level object view of queues and matching in the context of communication between nodes. Preferably, the Queue manipulation logic, Match Table manipulation logic and the Sequence Table manipulation logic are implemented in the adapter hardware which does not keep any state or resources in it that depend on the size of the system or the number of queues/tables instantiated. The actual states of these objects may be kept in the program memory, so that the adapter hardware is memoryless.

Abstract: A compiler implemented method performs call-site specific selective automatic procedure integration over a program call graph. Individual call sites are selected for inlining based on two cost functions without the need for profiling. A routine cost function models each procedure and attempts to capture the cost of inlining a procedure. A call site cost function models the region around each call site, and attempts to capture the benefit of inlining or cloning a particular call site. Inlining is performed only at call sites for which both the routine cost function is less than a first threshold value and the call site cost function is less than a second threshold value.

Abstract: A computer implemented system, two-step method and computer program product which improves the operations of multi-feature extraction and which efficiently develops classification rules from a large training database. Specifically, given a large training set of data tuples, the first phase, called the feature identification phase, identifies features, which have good power in separating data tuples, based on a subset of the training set. A feature that has a good power in correlating data tuples into groups is said to have a good discriminating power. Since the feature identification phase is performed on a subset of the training set, processing costs are minimized. Limiting this phase to the identification of features having good discriminating power ensures that the use of a subset of the training set does not adversely affect the validity of the conclusions drawn therefrom.

Abstract: An architecture and key algorithms are presented that simplify the implementation and improve the functionality of knowledge-based systems that employ case-based reasoning for data navigation tasks (e.g., solving help desk problems using a database of past problem instances). The various embodiments of the invention improve on the state-of-the-art in several respects: (a) provides greatly increased flexibility in choosing attributes relevant to a case in that choices are made dynamically at case acquisition instead of statically when the knowledge-based system is designed; (b) increased system flexibility and decreased system complexity by not requiring designers to specify the manner in which attribute values are compared; (c) simplified system design by avoiding the use of similarity metrics; (d) allows for the incremental acquisition of data in problem instances, instead of requiring that all potentially relevant data be collected; and (e) provides a way to explain why a problem solution was chosen.

Abstract: A system and method for Video-on-Demand (VOD) satisfying the desire of `surfers`. The process entails a server operating in a standard NVOD mode, whereby it continuously sends multiple copies of each program on separate channels. Each copy is delayed by a staggered time interval. The invention provides a way to fulfill a VOD user request asynchronous with the start of a NVOD transmission but which still makes primary use of the NVOD transmission for that requestor. The system and method combines independent server service satisfying an asynchronous request for only a short duration, usually less than or equal to the NVOD staggered time interval, and then aligns that requestor with an immediately previously commenced NVOD transmission.

Abstract: An efficient computer implemented method of mining path traversal patterns in a communications network. The method of the present invention comprises two steps. A method, called MF (standing for maximal forward references), is first used to convert an original sequence of log data into a set of traversal subsequences. Each traversal subsequence represents a maximal forward reference from the starting point of a user access. This step of converting the original log sequence into a set of maximal forward references will filter out the effect of backward references which are mainly made for ease of traveling, and enable us to concentrate on mining meaningful user access sequences. Accordingly, when backward references occur, a forward reference path terminates. This resulting forward reference path is termed a maximal forward reference. After a maximal forward reference is obtained, we back track to the starting point of the forward reference and begin a new forward reference path.

Abstract: A system and method for recovering from failures in the disk access path of a clustered computing system. Each node of the clustered computing system is provided with proxy software for handling physical disk access requests from applications executing on the node and for directing the disk access requests to an appropriate server to which the disk is physically attached. The proxy software on each node maintains state information for all pending requests originating from that node. In response to detection of a failure along the disk access path, the proxy software on all of the nodes directs all further requests for disk access to a secondary node physically attached to the same disk.

Abstract: Image analysis apparatus comprises a two dimensional array of photodetectors (80) for receiving light from pixels of a display screen (90). The intensity values generated by the photodetectors (80) in response to incident light from the pixels are stored in a memory (40). The photodetectors' spatial geometry is mapped onto the display screen's known pixel geometry. A processor (50) determines one or more performance parameters of the display screen (90) as a function of the intensity values stored in the memory (40) and of the photodetectors mapped spatial geometry (80). Calibration and correlation problems associated with prior art measurement techniques are avoided by mapping the spacing of the array of photodetectors onto the known geometry of the display pixel structure. Also, the apparatus can be conveniently incorporated into a handset and therefore does not require complicated positioning jigs in use.