Power companies can benefit from the use of knowledge discovery methods and statistical machine learning for preventive maintenance. We introduce a general process for transforming historical electrical grid data into models that aim to predict the risk of failures for components and systems. These models can be used directly by power companies to assist with prioritization of maintenance and repair work. Specialized versions of this process are used to produce 1) feeder failure rankings, 2) cable, joint, terminator, and transformer rankings, 3) feeder Mean Time Between Failure (MTBF) estimates, and 4) manhole events vulnerability rankings. The process in its most general form can handle diverse, noisy, sources that are historical (static), semi-real-time, or realtime, incorporates state-of-the-art machine learning algorithms for prioritization (supervised ranking or MTBF), and includes an evaluation of results via cross-validation and blind test. Above and beyond the ranked lists and MTBF estimates are business management interfaces that allow the prediction capability to be integrated directly into corporate planning and decision support; such interfaces rely on several important properties of our general modeling approach: that machine learning features are meaningful to domain experts, that the processing of data is transparent, and that prediction results are accurate enough to support sound decision making. We discuss the challenges in working with historical electrical grid data that were not designed for predictive purposes. The “rawness” of these data contrasts with the accuracy of the statistical models that can be obtained from the process; these models are sufficiently accurate to assist in maintaining New York City’s electrical grid.

Concern for the environment and energy security is changing the way we think about energy. Grid-enabled passenger vehicles, like electric vehicles (EV) and plug-in hybrid electric vehicles (PHEV) can help address environmental and energy issues. Automakers have recognized that electric drive vehicles are critical to the future of the industry. However, some challenges exist to greater adoption: the perception of cost, EV range, access to charging, potential impacts to the grid, and lack of public awareness about the availability and practicality of these vehicles. Although the current initial price for EV's is higher, their operating costs are lower. Policies that reduce the total cost of ownership of EVs and PHEVs, compared to conventional internal combustion engine (ICE) vehicles, will lead to faster market penetration. Greater access to charging infrastructure will also accelerate public adoption. Smart grid technology will optimize the vehicle integration with the grid, allowing intelligent and efficient use of energy. By coordinating efforts and using a systems perspective, the advantages of EVs and PHEVs can be achieved using the least resources. This paper analyzes these factors, their rate of acceleration and how they may synergistically align for the electrification of vehicles.

Automatic seizure detection is becoming popular in modern epilepsy monitoring units since it assists diagnostic monitoring and reduces manual review of large volumes of EEG recordings. In this paper, we describe the application of machine learning algorithms for building patient-specific seizure detectors on multiple frequency bands of intra-cranial electroencephalogram (iEEG) recorded by a dense Micro-Electrode Array (MEA). The MEA is capable of recording at a very high sampling rate (30 KHz) producing an avalanche of time series data. We explore subsets of this data to build seizure detectors - we discuss several methods for extracting univariate and bivariate features from the channels and study the effectiveness of using high dimensional clustering algorithms such as K-means and Subspace clustering for constructing the model. Future work involves design of more robust seizure detectors using other features and non-parametric clustering techniques, detection of artifacts and understanding the generalization properties of the models.

For many ranking applications we would like to understand not only which items are top-ranked, but also why they are top-ranked. However, many of the best ranking algorithms (e. g., SVMs) are black boxes that give little information about the factors for their rankings. We describe and demonstrate a new approach that can work in conjunction with any ranking algorithm to discover explanations for the items at the top of the rankings. These explanations are in the form of rules expressed as boolean combinations of attribute-value expressions. These rules are discovered by contrasting attributes of items drawn from both the top and bottom of a ranking list, looking for items that have high leverage, corresponding to rules with broad coverage and sharp differentiations. We include empirical results to demonstrate the utility of our method.

We describe a method for classifying news stories using Memory Based Reasoning (MBR) a k-nearest neighbor method), that does not require manual topic definitions. Using an already coded training database of about 50,000 stories from the Dow Jones Press Release News Wire, and SEEKER [Stanfill] (a text retrieval system that supports relevance feedback) as the underlying match engine, codes are assigned to new, unseen stories with a recall of about 80% and precision of about 70%. There are about 350 different codes to be assigned. Using a massively parallel supercomputer, we leverage the information already contained in the thousands of coded stories and are able to code a story in about 2 seconds. Given SEEKER, the text retrieval system, we achieved these results in about two person-months. We believe this approach is effective in reducing the development time to implement classification systems involving large number of topics for the purpose of classification, message routing etc.

We have developed a software system, PHI-PSI, on the Connection Machine that uses a parallel algorithm to retrieve and use information from a database of 112 known protein structures (selected from the Brookhaven Protein Databank) to predict the structures of other proteins. The φ and ψ angles of each amino acid (the angles each amino acid forms with its immediate neighbors) in a protein are used to represent its 3-D structure. PHI-PSI's algorithm is based on the idea of Memory-based reasoning (MBR) [10] and extends it to include a recursive procedure to refine its initial prediction and a “window” of varying sizes to look at different contexts of an input. PHI-PSI has been tested with all the available data. Initial results show that it performs better than distribution-based guesses for most of the φ and ψ angle values.

In this paper we present a parallel document ranking algorithm suitable for use on databases of 1-1000 GB, resident on primary or secondary storage. The algorithm is based on inverted indexes, and has two advantages over a previously published parallel algorithm for retrieval based on signature files. First, it permits the employment of ranking strategies which cannot be easily implemented using signature files, specifically methods which depend on document-term weighting. Second, it permits the interactive searching of databases resident on secondary storage. The algorithm is evaluated via a mixture of analytic and simulation techniques, with a particular focus on how cost-effectiveness and efficiency change as the size of the database, number of processors, and cost of memory are altered. In particular, we find that if the ratio of the number of processors and/or disks to the size of the database is held constant, then the cost-effectiveness of the resulting system remains constant. Furthermore, for a given size of database, there is a number of processors which optimizes cost-effectiveness. Estimated response times are also presented. Using these methods, it appears that cost-effective interactive access to databases in the 100-1000 GB range can be achieved using current technology.

The language of scene descriptions must allow a hearer to build structures of schemas similar (to some level of detail) to those the speaker has built via perceptual processes. The understanding process in general requires a hearer to create and run "event simulations" to check the consistency and plausibility of a "picture" constructed from a speaker's description. A speaker must also run similar event simulations on his own descriptions in order to be able to judge when the hearer has been given sufficient information to construct an appropriate "picture", and to be able to respond appropriately to the hearer's questions about or responses to the scene description.In this paper I explore some simple scene description examples in which a hearer must make judgements involving reasoning about scenes, space, common-sense physics, cause-effect relationships, etc. While I propose some mechanisms for dealing with such scene descriptions, my primary concern at this time is to flesh out our understanding of just what the mechanisms must accomplish: what information will be available to them and what information must be found or generated to account for the inferences we know are actually made.