Discovering user preferences by using time entries in click-through data to improve search engine results

The search engine log files have been used to gather direct user feedback on the relevancy of the documents presented in the results page. Typically the relative position of the clicks gathered from the log files is used a proxy for the direct user feedback. In this paper we identify reasons for the incompleteness of the relative position of clicks for deciphering the user preferences. Hence, we propose the use of time spent by the user in reading through the document as indicative of user preference for a document with respect to a query. Also, we identify the issues involved in using the time measure and propose means to address them.

Low PAPR Square STBCs from Complex Partial-Orthogonal Designs (CPODs)

Space-time codes from complex orthogonal designs (CODs) with no zero entries offer low Peak to Average Power Ratio (PAPR) and avoid the problem of switching off antennas. But square CODs for 2(a) antennas with a + 1. complex variables, with no zero entries were discovered only for a <= 3 and if a + 1 = 2(k), for k >= 4. In this paper, a method of obtaining no zero entry (NZE) square designs, called Complex Partial-Orthogonal Designs (CPODs), for 2(a+1) antennas whenever a certain type of NZE code exists for 2(a) antennas is presented. Then, starting from a so constructed NZE CPOD for n = 2(a+1) antennas, a construction procedure is given to obtain NZE CPODs for 2n antennas, successively. Compared to the CODs, CPODs have slightly more ML decoding complexity for rectangular QAM constellations and the same ML decoding complexity for other complex constellations. Using the recently constructed NZE CODs for 8 antennas our method leads to NZE CPODs for 16 antennas. The class of CPODs do not offer full-diversity for all complex constellations. For the NZE CPODs presented in the paper, conditions on the signal sets which will guarantee full-diversity are identified. Simulation results show that bit error performance of our codes is same as that of the CODs under average power constraint and superior to CODs under peak power constraint.

Power dissipation for systems with junctions of multiple quantum wires

We study power dissipation for systems of multiple quantum wires meeting at a junction, in terms of a current splitting matrix (M) describing the junction. We present a unified framework for studying dissipation for wires with either interacting electrons (i.e., Tomonaga-Luttinger liquid wires with Fermi-liquid leads) or noninteracting electrons. We show that for a given matrix M, the eigenvalues of (MM)-M-T characterize the dissipation, and the eigenvectors identify the combinations of bias voltages which need to be applied to the different wires in order to maximize the dissipation associated with the junction. We use our analysis to propose and study some microscopic models of a dissipative junction which employ the edge states of a quantum Hall liquid. These models realize some specific forms of the M matrix whose entries depends on the tunneling amplitudes between the different edges.

Bounding Run-Times of Local Adiabatic Algorithms

A common trick for designing faster quantum adiabatic algorithms is to apply the adiabaticity condition locally at every instant. However it is often difficult to determine the instantaneous gap between the lowest two eigenvalues, which is an essential ingredient in the adiabaticity condition. In this paper we present a simple linear algebraic technique for obtaining a lower bound on the instantaneous gap even in such a situation. As an illustration, we investigate the adiabatic un-ordered search of van Dam et al. [17] and Roland and Cerf [15] when the non-zero entries of the diagonal final Hamiltonian are perturbed by a polynomial (in log N, where N is the length of the unordered list) amount. We use our technique to derive a bound on the running time of a local adiabatic schedule in terms of the minimum gap between the lowest two eigenvalues.

Low PAPR STBCs from Complex Partial-Orthogonal Designs (CPODs)

Space-time codes from complex orthogonal designs (CODs) with no zero entries offer low Peak to Average power ratio (PAPR) and avoid the problem of turning off antennas. But CODs for 2(a) antennas with a + 1 complex variables, with no zero entries are not known in the literature for a >= 4. In this paper, a method of obtaining no zero entry (NZE) codes, called Complex Partial-Orthogonal Designs (CPODs), for 2(a+1) antennas whenever a certain type of NZE code exists for 2(a) antennas is presented. This is achieved with slight increase in the ML decoding complexity for regular QAM constellations and no increase for other complex constellations. Since NZE CODs have been constructed recently for 8 antennas our method leads to NZE CPODs for 16 antennas. Moreover, starting from certain NZE CPODs for n antennas, a construction procedure is given to obtain NZE CPODs for 2n antennas. The class of CPODs do not offer full-diversity for all complex constellations. For the NZE CPODs presented in the paper, conditions on the signal sets which will guarantee full-diversity are identified. Simulations results show that bit error performance of our codes under average power constraint is same as that of the CODs and superior to CODs under peak power constraint.

Complex near-orthogonal designs with no zero entry

Zero entries in complex orthogonal designs (CODs) impede their practical implementation. In this paper, a method of obtaining a no zero entry (NZE) code for 2(k+1) antennas whenever a NZE code exists for 2(k) antennas is presented. This is achieved with slight increase in the ML decoding complexity for regular QAM constellations and no increase for other complex constellations. Since NZE CODs have been constructed recently for 8 antennas our method leads to NZE codes for 16 antennas. Simulation results show good performance of our new codes compared to the well known constructions for 16 and 32 antennas under peak power constraints.

On the maximal rate of non-square STBCs from complex orthogonal designs

A Linear Processing Complex Orthogonal Design (LPCOD) is a p x n matrix epsilon, (p >= n) in k complex indeterminates x(1), x(2),..., x(k) such that (i) the entries of epsilon are complex linear combinations of 0, +/- x(i), i = 1,..., k and their conjugates, (ii) epsilon(H)epsilon = D, where epsilon(H) is the Hermitian (conjugate transpose) of epsilon and D is a diagonal matrix with the (i, i)-th diagonal element of the form l(1)((i))vertical bar x(1)vertical bar(2) + l(2)((i))vertical bar x(2)vertical bar(2)+...+ l(k)((i))vertical bar x(k)vertical bar(2) where l(j)((i)), i = 1, 2,..., n, j = 1, 2,...,k are strictly positive real numbers and the condition l(1)((i)) = l(2)((i)) = ... = l(k)((i)), called the equal-weights condition, holds for all values of i. For square designs it is known. that whenever a LPCOD exists without the equal-weights condition satisfied then there exists another LPCOD with identical parameters with l(1)((i)) = l(2)((i)) = ... = l(k)((i)) = 1. This implies that the maximum possible rate for square LPCODs without the equal-weights condition is the same as that or square LPCODs with equal-weights condition. In this paper, this result is extended to a subclass of non-square LPCODs. It is shown that, a set of sufficient conditions is identified such that whenever a non-square (p > n) LPCOD satisfies these sufficient conditions and do not satisfy the equal-weights condition, then there exists another LPCOD with the same parameters n, k and p in the same complex indeterminates with l(1)((i)) = l(2)((i)) = ... = l(k)((i)) = 1.

Intent Based Clustering of Search Engine Query Log

The keyword based search technique suffers from the problem of synonymic and polysemic queries. Current approaches address only theproblem of synonymic queries in which different queries might have the same information requirement. But the problem of polysemic queries,i.e., same query having different intentions, still remains unaddressed. In this paper, we propose the notion of intent clusters, the members of which will have the same intention. We develop a clustering algorithm that uses the user session information in query logs in addition to query URL entries to identify cluster of queries having the same intention. The proposed approach has been studied through case examples from the actual log data from AOL, and the clustering algorithm is shown to be successful in discerning the user intentions.

On the maximal rate of (n+1) x n and (n+2) x n complex orthogonal designs

For p x n complex orthogonal designs in k variables, where p is the number of channels uses and n is the number of transmit antennas, the maximal rate L of the design is asymptotically half as n increases. But, for such maximal rate codes, the decoding delay p increases exponentially. To control the delay, if we put the restriction that p = n, i.e., consider only the square designs, then, the rate decreases exponentially as n increases. This necessitates the study of the maximal rate of the designs with restrictions of the form p = n+1, p = n+2, p = n+3 etc. In this paper, we study the maximal rate of complex orthogonal designs with the restrictions p = n+1 and p = n+2. We derive upper and lower bounds for the maximal rate for p = n+1 and p = n+2. Also for the case of p = n+1, we show that if the orthogonal design admit only the variables, their negatives and multiples of these by root-1 and zeros as the entries of the matrix (other complex linear combinations are not allowed), then the maximal rate always equals the lower bound.

A numerical study of division of flow in open channels

A two-dimensional numerical model which employs the depth-averaged forms of continuity and momentum equations along with k-e turbulence closure scheme is used to simulate the flow at the open channel divisions. The model is generalised to flows of arbitrary geometries and MacCormack finite volume method is used for solving governing equations. Application of cartesian version of the model to analyse the flow at right-angled junction is presented. The numerical predictions are compared with experimental data of earlier investigators and measurements made as part of the present study. Performance of the model in predicting discharge distribution, surface profiles, separation zone parameters and energy losses is evaluated and discussed in detail. To illustrate the application of the numerical model to analyse the flow in acute angled offtakes and streamlined branch entries, a few computational results are presented.

Robust Formulations for Handling Uncertainty in Kernel Matrices

We study the problem of uncertainty in the entries of the Kernel matrix, arising in SVM formulation. Using Chance Constraint Programming and a novel large deviation inequality we derive a formulation which is robust to such noise. The resulting formulation applies when the noise is Gaussian, or has finite support. The formulation in general is non-convex, but in several cases of interest it reduces to a convex program. The problem of uncertainty in kernel matrix is motivated from the real world problem of classifying proteins when the structures are provided with some uncertainty. The formulation derived here naturally incorporates such uncertainty in a principled manner leading to significant improvements over the state of the art. 1.

Dynamic matrix rank with partial lookahead

We consider the problem of maintaining information about the rank of a matrix $M$ under changes to its entries. For an $n \times n$ matrix $M$, we show an amortized upper bound of $O(n^{\omega-1})$ arithmetic operations per change for this problem, where $\omega < 2.376$ is the exponent for matrix multiplication, under the assumption that there is a {\em lookahead} of up to $\Theta(n)$ locations. That is, we know up to the next $\Theta(n)$ locations $(i_1,j_1),(i_2,j_2),\ldots,$ whose entries are going to change, in advance; however we do not know the new entries in these locations in advance. We get the new entries in these locations in a dynamic manner.

mRegistry: a registry representation for fault diagnosis

Microsoft Windows uses the notion of registry to store all configuration information. The registry entries have associations and dependencies. For example, the paths to executables may be relative to some home directories. The registry being designed with faster access as one of the objectives does not explicitly capture these relations. In this paper, we explore a representation that captures the dependencies more explicitly using shared and unifying variables. This representation, called mRegistry exploits the tree-structured hierarchical nature of the registry, is concept-based and obtained in multiple stages. mRegistry captures intra-block, inter-block and ancestor-children dependencies (all leaf entries of a parent key in a registry put together as an entity constitute a block thereby making the block as the only child of the parent). In addition, it learns the generalized concepts of dependencies in the form of rules. We show that mRegistry has several applications: fault diagnosis, prediction, comparison, compression etc.

Tractable theories for the synthesis of neural networks

The Radius of Direct attraction of a discrete neural network is a measure of stability of the network. it is known that Hopfield networks designed using Hebb's Rule have a radius of direct attraction of Omega(n/p) where n is the size of the input patterns and p is the number of them. This lower bound is tight if p is no larger than 4. We construct a family of such networks with radius of direct attraction Omega(n/root plog p), for any p greater than or equal to 5. The techniques used to prove the result led us to the first polynomial-time algorithm for designing a neural network with maximum radius of direct attraction around arbitrary input patterns. The optimal synaptic matrix is computed using the ellipsoid method of linear programming in conjunction with an efficient separation oracle. Restrictions of symmetry and non-negative diagonal entries in the synaptic matrix can be accommodated within this scheme.

Role of water contamination within the GC column of a GasBench II peripheral on the reproducibility of O-18/O-16 ratios in water samples

The GasBench II peripheral along with MAT 253 combination provides a more sensitive platform for the determination of water isotope ratios. Here, we examined the role of adsorbed moisture within the gas chromatography (GC) column of the GasBench II on measurement uncertainties. The uncertainty in O-18/O-16 ratio measurements is determined by several factors, including the presence of water in the GC. The contamination of GC with water originating from samples as water vapour over a longer timeframe is a critical factor in determining the reproducibility of O-18/O-16 ratios in water samples. The shift in isotope ratios observed in the experiment under dry and wet conditions correlates strongly with the retention time of analyte CO2, indicating the effect of accumulated moisture. Two possible methods to circumvent or minimise the effect of adsorbed water on isotope ratios are presented here. The proposed methodology includes either the regular baking of the GC column at a higher temperature (120 degrees C) after analysis of a batch of 32 sample entries or conducting the experiment at a low GC column temperature (22.5 degrees C). The effects of water contamination on long-term reproducibility of reference water, with and without baking protocol, have been described.