Abstract: The features that are presented to an evolutionary algorithm are preprocessed to generate combination features that may be more efficient in distinguishing among classifications than the individual features that comprise the combination feature. An initial set of features is defined that includes a large number of potential features, including the generated features that are combinations of other features. These features include, for example, all of the words used in a collection of content material that has been previously classified, as well as combination features based on these features, such as all the noun and verb phrases used. This pool of original features and combination features are provided to an evolutionary algorithm for a subsequent evaluation, generation, and determination of the best subset of features to use for classification.

Abstract: A method and apparatus are disclosed for displaying available television programs with an indication of the strength of the contribution of one or more program attributes or features to the overall recommendation score assigned by a program recommender. The program and corresponding indication of the strength of the contribution of one or more program attributes can be presented to the user, for example, using grids listing the available television programs by time and date, channel and title. The overall recommendation scores or component scores associated with each program are also displayed to the user. The overall recommendation scores or component scores can be displayed with each program directly or can be mapped onto a color spectrum or another visual cue, such as a variable size-of-text, rate of blinking or bar height. The visual cues are then applied to each program in the program grid in accordance with the present invention.

Abstract: A method and apparatus for validating recommendations generated by a television program recommender uses programmed viewing agents, in which a viewing agent is programmed with a set of rules that characterize the viewing preferences of a modeled viewer. During a training phase, the programmed rules of a viewing agent are applied to a set of training programs to obtain an agent viewing history, which is processed by a profiler to derive an agent profile containing a set of inferred rules. During an evaluation phase, the programmed rules of the viewing agent are applied to test programs to obtain an agent evaluation viewing set. In parallel, the television program recommender generates a set of program recommendations by applying the agent profile to the test programs. The agent evaluation viewing set is then compared with the program recommendations.

Abstract: A method and apparatus are disclosed for validating recommendations generated by a television program recommender using programmed viewing agents. A viewing agent is programmed with a set of rules that characterize the viewing preferences of a modeled viewer. During a training phase, the programmed rules of a viewing agent are applied to a set of training programs to obtain an agent viewing history. The method and apparatus estimates the required size of the viewing history to provide a given level of accuracy. The viewing agents can be programmed to introduce one or more random shows into the viewing history, or to change the preferences of the viewing agent over time, thereby allowing the television program recommender validator to evaluate how the program recommender processes such non-stationaries.

Abstract: A method and apparatus for displaying available television programs with an indication of the recommendation score assigned to each program by a television programming recommender, includes a television programming recommender for evaluating each of the programs in an electronic programming guide (EPG) in a conventional manner to identify programs of interest to a particular user. An indication of the numerical recommendation scores associated with each program are also displayed to the user, for example, using program grids listing the available television programs. The numerical recommendation scores can be displayed with each program directly or can be mapped onto a color spectrum or another visual cue, such as a variable size-of-text or rate of blinking, that permits the user to quickly locate programs of interest. Television channels can be sorted in the program grid according to a normalized recommendation score for the time period being examined.

Abstract: A method and system for controlling the growth of a features frequency profile of a time-ordered sequence of events, wherein each event has features specific to each event. The events are sequentially processed in an order of processing. The processing includes selecting for each event processed at least one feature comprised by the event. The processing updates a frequency count of each feature selected. The frequency counts are periodically reduced in magnitude by a reduction factor. Frequency counts are selected for deletion upon satisfaction of a condition that favors deletion of those frequency counts having a magnitude less than a threshold value. The selected frequency counts are then deleted. The present invention employs an economical use of memory to store data associated with the features frequency profile and uses a features preference profile that is more responsive to recent information than to older information.

Abstract: A modular optimizer, for use in determining a configuration of a production line with one or more component placement machines, is configured to handle precedence constraints. The precedence constraints may be of the form A B MT, which specifies that part A must be placed on a designated assembly board before part B if either part is to be placed by a machine type MT. A given set of precedence constraints includes at least a first class of constraints that apply to only one component placement machine type and a second class of constraints that apply to more than one component placement machine type. Assignment of constraints to the different classes is based on decisions of a part splitter module of the modular optimizer regarding which parts are assigned to which machine types. Each of the constraints in the first class of constraints associated with a given machine type are handled in a corresponding machine module of the modular optimizer.

Abstract: Neural network architectures are represented by symbol strings. An initial population of networks is trained and evaluated. The strings representing the fittest networks are modified according to a genetic algorithm and the process is repeated until an optimized network is produced.

Abstract: Neural network architectures are represented by symbol strings. An initial population of networks is trained and evaluated. The strings representing the fittest networks are modified according to a genetic algorithm and the process is repeated until an optimized network is produced.

Abstract: A modular optimizer, for use in determining a configuration of a production line with one or more component placement machines, is configured to learn a foreign optimizer module associated with a component placement machine type foreign to the modular optimizer. The modular optimizer includes an adapting estimator which estimates an output value, such as a placement cycle time measure, for the foreign optimizer module. The estimated output value may be used in determining the configuration of the production line.

Abstract: A method for optimizing the performance of an algorithm for detecting predetermined content in a media information stream, and a program and apparatus that operate in accordance with the method. The algorithm is a function of a set of parameters. The method comprises the steps of performing the algorithm at least once to detect the predetermined content in the media information stream, while employing a respective set of parameters in the algorithm for each performance thereof, and automatically evolving at least one respective set of parameters employed in the algorithm to maximize the degree of accuracy at which the algorithm detects the predetermined content in the media information stream.

Abstract: A method and system for providing hierarchical decision fusion of recommender scores, wherein at least two levels of fusion are provided. In a method, a plurality of recommenders at a first level are grouped according to topics of interest. A plurality of first level fusion centers receive a number of outputs from a predetermined number of recommenders. The first level fusion centers output a first enhanced decision level, and a series of second level fusion centers receive a predetermined number of the first enhanced decision, and a second fusing step occurs to result in a second enhanced decision level. The groups can be reading history, music, viewing history, purchasing history, and can be intermixed, so that the enhanced decision may recommend a particular movie based on both the ranking about movies and music.

Abstract: The noise associated with conventional techniques for evolutionary improvement of neural network architectures is reduced so that of an optimum architecture can be determined more efficiently and more effectively. Parameters that affect the initialization of a neural network architecture are included within the encoding that is used by an evolutionary algorithm to optimize the neural network architecture. The example initialization parameters include an encoding that determines the initial nodal weights used in each architecture at the commencement of the training cycle. By including the initialization parameters within the encoding used by the evolutionary algorithm, the initialization parameters that have a positive effect on the performance of the resultant evolved network architecture are propagated and potentially improved from generation to generation. Conversely, initialization parameters that, for example, cause the resultant evolved network to be poorly trained, will not be propagated.

Abstract: A recommendation system and method are disclosed. In the system and method, the personal schedule of the user is used to modify the recommendation functions of media events. The personal schedule may be entered by the user or determined through monitoring over time. An exemplary recommendation function modification is if a media event ends after the user's bedtime, as indicated by the personal schedule. In this example, the recommendation function of that event will be reduced in value because the user will likely go to bed before the event is over.

Abstract: A system employing an automated collaborative filtering process for recommending an item to a viewer based upon feedback data, implicit data, and/or explicit data corresponding to a primary viewer as well as secondary viewers is disclosed. A first act of the automated collaborative filtering process is to match data indicative of a viewing of a first group of items by the primary viewer to data indicative of a viewing of a second group of items by the secondary viewers. A second act of the automated collaborative filtering process is to generate a recommendation of the item by the primary viewer as a function of data indicative of one or more attributes of the item as compared to the data matching accomplished in the first act.

Abstract: A program recommendation system employing a program record module and one of various nearest neighbor modules is disclosed. In response to a reception of a program record, the program record module converts each key field of the program record into a feature value. A single neighbor module selectively generates a recommendation of a program corresponding to the program record based upon a stored program record qualifying as a nearest neighbor of the received program record. A multiple neighbor module selectively generates a recommendation of the program corresponding to the program record based upon N number of stored program records qualifying as N number of nearest neighbors of the received program record. A neighbor cluster selectively generates a recommendation of the program corresponding to the program record based upon the cluster of stored program records qualifying as the nearest neighbor of the received program record.

Abstract: A security system incorporating a reasoning system and security rules and processes. Transponders may be triggered and sensed from a distance to identify both items and individuals. These sensed identifiers are processed by the reasoning system to determine whether each identified item is authorized to be removed from or brought into a secured location by the identified individual. The system modifies and optimizes its rules and processes based on assessments of security events. The security system enforces these security rules and receives feedback from authorized security personnel. A learning system is configured to modify existing rules or create new rules in conformance with the feedback from the authorized security personnel.

Abstract: A genetic algorithm is used to balance a line of pick and place machines. The genetic algorithm uses a modularized chromosome string having at least three parts indicating 1) a division of parts between the pick and place machines, 2) a layout of a first pick and place machine, and 3) a layout of a second pick and place machine, respectively. A heuristic layout generator cycles with the genetic algorithm to create simulated layouts from populations of chromosome strings produced by the genetic algorithm. The heuristic layout generator is also modularized, having separate modules corresponding to the three parts of the chromosome string.

Abstract: sets or evaluation on sets among the evaluated neural networks may cause potentially worthwhile architectures to be rejected prematurely, obviating the advantages realizable by a directed trial and error process. It is an object of this invention to provide a method for improving neural network architectures via an evolutionary algorithm that reduces the adverse effects of the noise that is introduced by the network initialization process. It is a further object of this invention to reduce the noise that is introduced by the network initialization process. It is a further object of this invention to provide an optimized network initialization process. It is a further object of this invention to reduce the noise that is introduced by the use of randomly selected training or evaluation input sets.

Abstract: An evolutionary algorithm evolves alternative architectures and parameters for an image classification system. In a preferred embodiment, a learning system is employed, and during the training period of the learning system, the architecture of the learning system is evolved so as to create a learning system that is well suited to the particular classification problem set. In like manner, other parameters of the image classification system are evolved by the evolutionary algorithm, including those that effect image characterization, learning, and classification. An initial set of parameters and architectures are used to create a set of trial classification systems. A number of pre-classified evaluation images are then applied to each system, and each system's resultant classifications for each test case is compared to the proper classification of each test case. Subsequent trial classification systems are evolved based upon the parameters and architecture of the better performing classification systems.