Syam Menon

Maximizing Accuracy of Shared Databases when Concealing Sensitive Patterns

The sharing of databases either within or across organizations raises the possibility of unintentionally revealing sensitive relationships contained in them. Recent advances in data-mining technology have increased the chances of such disclosure. Consequently, firms that share their databases might choose to hide these sensitive relationships prior to sharing. Ideally, the approach used to hide relationships should be impervious to as many data-mining techniques as possible, while minimizing the resulting distortion to the database. This paper focuses on frequent item sets, the identification of which forms a critical initial step in a variety of data-mining tasks. It presents an optimal approach for hiding sensitive item sets, while keeping the number of modified transactions to a minimum. The approach is particularly attractive as it easily handles databases with millions of transactions. Results from extensive tests conducted on publicly available real data and data generated using IBMs synthetic data generator indicate that the approach presented is very effective, optimally solving problems involving millions of transactions in a few seconds.

Assigning cells to switches in cellular networks by incorporating a pricing mechanism into Simulated annealing

Handoff and cabling costs play key roles in the design of cellular telecommunications networks. Efficient homing of cells to switches can have a significant impact on each of these costs. In the presence of capacity constraints at the switches, the problem of assigning cells to switches becomes a difficult one to solve, with all effective solution approaches being based on heuristic techniques. This paper presents a hybrid heuristic which integrates ideas from linear programming into a simulated annealing framework. Extensive computational results are presented comparing the performance of the heuristic with the lower bound obtained from the linear programming relaxation. These results indicate that this price influenced simulated annealing (PISA) procedure is extremely efficient, usually providing solutions with gaps around 1% in less than 1 s.

Efficient scheduling of Internet banner advertisements

Despite the slowdown in the economy, advertisement revenue remains a significant source of income for many Internet-based organizations. Banner advertisements form a critical component of this income, accounting for 40 to 50 percent of the total revenue. There are considerable gains to be realized through the efficient scheduling of banner advertisements. This problem has been observed to be intractable via traditional optimization techniques, and has received only limited attention in the literature. This paper presents a procedure to generate advertisement schedules under the most commonly used advertisement pricing scheme---the CPM model. The solution approach is based on Lagrangean decomposition and is seen to provide extremely good advertisement schedules in a relatively short period of time, taking only a few hundred seconds of elapsed time on a 450 MHz PC compared to a few thousand seconds of CPU time on a workstation that other approaches need. Additionally, this approach can be incorporated into an actual implementation with minimal alterations and hence is of particular interest.

Allocating fragments in distributed databases

For a distributed database system to function efficiently, the fragments of the database need to be located, judiciously at various sites across the relevant communications network. The problem of allocating these fragments to the most appropriate sites is a difficult one to solve, however, with most approaches available relying on heuristic techniques. Optimal approaches are usually based on mathematical programming, and formulations available for this problem are based on the linearization of nonlinear binary integer programs and have been observed to be ineffective except on very small problems. This paper presents new integer programming formulations for the nonredundant version of the fragment allocation problem. This formulation is extended to address problems which have both storage and processing capacity constraints; the approach is observed to be particularly effective in the presence of capacity restrictions. Extensive computational tests conducted over a variety of parameter values indicate that the reformulations are very effective even on relatively large problems, thereby reducing the need for heuristic approaches.

Order Allocation for Stock Cutting in the Paper Industry

A common problem encountered in paper-production facilities is that of allocating customer orders to machines so as to minimize the total cost of production. It can be formulated as adual-angular integer program, with identical machines inducing symmetry. While the potential advantages of decomposing large mathematical programs into smaller subproblems have long been recognized, the solution of decomposableinteger programs remains extremely difficult. Symmetry intensifies the difficulty. This paper develops an approach, based on the construction of tight subproblem bounds, to solve decomposable dual-angular integer programs and successfully applies it to solve the problem from the paper industry. This method is of particular interest as it significantly reduces the impact of symmetry.

Minimizing Information Loss and Preserving Privacy

The need to hide sensitive information before sharing databases has long been recognized. In the context of data mining, sensitive information often takes the form of itemsets that need to be suppressed before the data is released. This paper considers the problem of minimizing the number of nonsensitive itemsets lost while concealing sensitive ones. It is shown to be an intractably large version of an NP-hard problem. Consequently, a two-phased procedure that involves the solution of two smaller NP-hard problems is proposed as a practical and effective alternative. In the first phase, a procedure to solve a sanitization problem identifies how the support for sensitive itemsets could be eliminated from a specific transaction by removing the fewest number of items from it. This leads to a modified frequent itemset hiding problem, where transactions to be sanitized are selected such that the number of nonsensitive itemsets lost, while concealing sensitive ones, is minimized. Heuristic procedures are developed for these problems using intuition derived from their integer programming formulations. Results from computational experiments conducted on a publicly available retail data set and three large data sets generated using IBMs synthetic data generator indicate that these approaches are very effective, solving problems involving up to 10 million transactions in a short period of time. The results also show that the process of sanitization has considerable bearing on the quality of solutions obtained.

Scheduling Banner Advertisements on the Web

Despite the slowdown in the economy, advertisement revenue remains a significant source of income for many Internet-based organizations. Banner advertisements form a critical component of this income, accounting for 40 to 50% of the total revenue. There are considerable gains to be realized through the efficient scheduling of banner advertisements. This problem has received only limited attention in the literature. This paper introduces formulations for an important version of the banner advertisement scheduling problem. Two solution approaches, one based on Lagrangean decomposition and the other based on column generation, are presented, along with extensive results based on 1,500 randomly generated data sets. These results suggest that, while both approaches do very well in general, column generation consistently performs better than Lagrangean decomposition, giving gaps of 0.0004% or better in 4.4 seconds or less, even on relatively large problem instances.

A sequential approach for optimal broadcast scheduling in packet radio networks

An important problem that arises in the design of packet radio networks is that of scheduling access to the high speed communications channel in such a way as to avoid interference while keeping the frame length to a minimum. The broadcast scheduling problem is known to be NP-hard and to date, this problem has been formulated as a nonlinear discrete optimization problem for a given frame length, and solved via heuristic approaches by parametrically varying the length of the frame. This paper presents a linear integer programming formulation for the composite problem of maximizing channel utilization while minimizing the length of the frame. It then introduces a solution approach based on solving two relatively easier (though still NP-complete) integer programs in succession. Computational experiments are conducted on a set of benchmark cases and additional randomly generated problem instances. Results show that this sequential integer programming approach is very effective, solving all the problems optimally within a few seconds. These results imply that optimal solutions can be identified in very little time for problems of realistic size, and that heuristic approaches will be needed only when problems get much larger than those considered in the literature to date.

Analyzing Sharing in Peer-to-Peer Networks Under Various Congestion Measures

Historically, the use of peer-to-peer (P2P) networks has been limited primarily to user-initiated exchanges of (mostly music) files over the Internet. This traditional view of P2P networks is changing, however, and the use of P2P networks has been suggested for delivering general-purpose content over the Web (or corporate intranets), even in real time. We analyze sharing in a P2P community in this new context under three different congestion measures: delay, jitter, and packet loss. Sharing is important to study in the presence of congestion because most existing research on P2P networks views congestion in the network as a relatively insignificant criterion. However, when delivering general-purpose content, congestion and its relationship to sharing is a critical factor that influences end-user performance. This paper looks at P2P networks from this new perspective by explicitly considering the effects of congestion on user incentives for sharing. We also propose a simple incentive mechanism that induces socially optimal sharing.

Effective reformulations for task allocation in distributed systems with a large number of communicating tasks

In any distributed processing environment, decisions need to be made concerning the assignment of computational task modules to various processors. Many versions of the task allocation problem have appeared in the literature. Intertask communication makes the assignment decision difficult; capacity limitations at the processors increase the difficulty. This problem is naturally formulated as a nonlinear integer program, but can be linearized to take advantage of commercial integer programming solvers. While traditional approaches to linearizing the problem perform well when only a few tasks communicate, they have considerable difficulty solving problems involving a large number of intercommunicating tasks. This paper introduces new mixed integer formulations for three variations of the task allocation problem. Results from extensive computational tests conducted over real and generated data indicate that the reformulations are particularly efficient when a large number of tasks communicate, solving reasonablylarge problems faster than other exact approaches available.