Critical dimension is the minimum number of features that is required to ensure the performance of a learning machine to be “high”. This critical dimension is usually unique to the learning machine and the ranking algorithm combination. Medical- and bio-informatics datasets are different from most other datasets in that there is an imbalance in most of these datasets and a high prediction accuracy often depends upon not just the overall accuracy but also the true positive and the false negative rates. To find a medically and bio-informatically accurate critical dimension and for better analysis of such datasets we develop two evaluation models, one using all features and the other using critical number of features. The performance measurements such as accuracy, specificity, sensitivity, area under the curve, F-score and kappa values are compared. This paper shows that at the critical dimension the evaluation model shows good results for all performance measurements measured on most datasets studied. The difference in performance measurements obtained using only critical number and using all features is significantly less, i.e., there is not much difference in sensitivity, specificity and other measurements calculated.

We present a classification algorithm built on our adaptation of the Generalized Lotka-Volterra model, well-known in mathematical ecology. The training algorithm itself consists only of computing several scalars, per each training vector, using a single global user parameter and then solving a linear system of equations. Construction of the system matrix is driven by our model and based on kernel functions. The model allows an interesting point of view of kernels' role in the inductive learning process. We describe the model through axiomatic postulates. Finally, we present the results of the preliminary validation experiments.

Although the paperless office has been imminent for decades, documents in paper form continue to be used extensively in almost all organizations. Present-day information systems are designed on the premise that any paper document in use will be either converted into electronic form or merely printed from electronic file(s) accessible to the system. Yet, paper is the medium of choice in many situations, mainly owing to its portability and usability, and the medium of necessity in others, especially where external communication or the traditional notion of authenticity are involved. Humans who find unique attractive features in both paper and electronic forms of documents, must survive this tension between the de-jure banishment of paper and its de-facto prevalence. In this paper, we propose to make paper documents first-class citizens by including them in the model underlying the information system. Specifically, we extend the schema of a document database with the notion of paper documents, physical locations, and the organizational hierarchy. This leads to an overall enhancement of document integrity and the ability to answer queries such as "where are the customer complaint letters we have received today?" and "which documents are in this filing cabinet?". Recent technological advances such as sensors have made the implementation of such a model very realistic.

As mobile devices get ubiquitous and grow in computational power, their management of interdependent data also becomes increasingly important. The mobile environment exhibits all the characteristics of a distributed database plus the feature of whimsical connectivity. Consequently, transactions respecting data consistency can suffer unbounded and unpredictable delays at both mobile and stationary nodes. The currently popular multi-tier model, in which mobile devices are in one end and always-connected stationary servers in the other, has certain practical advantages. However, it assumes that all integrity constraints are evaluated at the servers and hence relies on the semantics of operations for any autonomy enhancement of the mobile devices. In this paper, we examine the idea of constraint localization in cases where two mobile nodes each own data that share a constraint. It relies on reformulation of a constraint into more flexible local constraints that give more autonomy to the mobile nodes. The scheme also involves dynamic changes of these local constraints through negotiation, which we call re-localization. To overcome the problem of simultaneous requests for such re-localization, we give algorithms along with experimental results indicating their effectiveness.

Some information in databases and knowledge bases often needs to be protected from disclosure to certain users. Traditional solutions involving multi-level mechanisms are threatened by the user's ability to infer higher level information from the semantics of the application. We concentrate on the revelation of secrets through a user running transactions in the presence of database integrity constraints. We develop a method of specifying secrets formally that not only exposes a useful structure and equivalence among secrets but also allows a theorem prover to detect certain security lapses during transaction compilation time.

An analysis of database transactions in the presence of database integrity constraints can lead to several modes of feedback to transaction designers. The different kinds of feedback include tests and updates that could be added to the transaction to make it obey the integrity constraints, as well as predicates representing post-conditions guaranteed by a transaction's execution. We discuss the various modes, meanings, and uses of feedback. We also discuss methods of generating feedback from integrity constraints, transaction details and theorems constituting both generic knowledge of database systems and specific knowledge about a particular database. Our methods are based on a running system that generates tailored theories about database systems from their schemas and uses these theories to prove that transactions obey integrity constraints.