Personalized web service recommendation through mishmash technique and deep learning model

Multimedia Tools and Applications - Discovering the relevant web services for specific applications in the dynamically changing business world becomes very critical. Researchers have used many...     

Enhanced layered segment trees: a pragmatic data structure for real-time processing of geometric objects

The data structure that is probably most used in the pattern recognition and image processing of geometric objects is the segment tree and its optimized variant, the “layered segment tree”. In all the versions currently known, except the work of Vaishnavi and Wood described later, these data structures do not operate in real time. Even in the latter scheme, although the structure can be implemented in real time and in an on-line fashion, the operation of insertion involves the sorting of the representations of the line segments in the tree. In essence, for all the reported algorithms, there is no known strategy to insert the segments one by one, other than the trivial strategy of processing them all together as in a batched-mode. In this paper, we present a strategy in which all the operations done on the tree can be done efficiently. Indeed, by improving the bottleneck, we prove that an arbitrary horizontal segment can be inserted into this data structure without invoking an expensive sorting process. We show that while this is accomplished by maintaining the same space and query complexity of the best-known algorithm, the version presented here is applicable to on-line real-time processing of line segments. The paper thus has applications in all areas of pattern recognition and image processing involving geometric objects.     

Recent Advances in Hierarchical Reinforcement Learning

Reinforcement learning is bedeviled by the curse of dimensionality: the number of parameters to be learned grows exponentially with the size of any compact encoding of a state. Recent attempts to combat the curse of dimensionality have turned to principled ways of exploiting temporal abstraction, where decisions are not required at each step, but rather invoke the execution of temporally-extended activities which follow their own policies until termination. This leads naturally to hierarchical control architectures and associated learning algorithms. We review several approaches to temporal abstraction and hierarchical organization that machine learning researchers have recently developed. Common to these approaches is a reliance on the theory of semi-Markov decision processes, which we emphasize in our review. We then discuss extensions of these ideas to concurrent activities, multiagent coordination, and hierarchical memory for addressing partial observability. Concluding remarks address open challenges facing the further development of reinforcement learning in a hierarchical setting.     

Spectral characterization and ASTER-based lithological mapping of an ophiolite complex: A case study from Neyriz ophiolite, SW Iran

The Neyriz ophiolite occurs along the Zagros suture zone in SW Iran, and is part of a 3000-km obduction belt thrusting over the edge of the Arabian continent during the late Cretaceous. This complex typically consists of altered dunites and peridotites, layered and massive gabbros, sheeted dykes and pillow lavas, and a thick sequence of radiolarites. Reflectance and emittance spectra of Neyriz ophiolite rock samples were measured in the laboratory and their spectra were used as endmembers in a spectral feature fitting (SFF) algorithm. Laboratory spectral reflectance measurements of field samples showed that in the visible through shortwave infrared (VNIR-SWIR) wavelength region the ultramafic and gabbroic rocks are characterized by ferrous-iron and Fe, MgOH spectral features, and the pillow lavas and radiolarites are characterized by spectral features of ferric-iron and AlOH. The laboratory spectral emittance spectra also revealed a wide wavelength range of SiO spectral features for the ophiolite rock units. After continuum removal of the spectra, the SFF classification method was applied to the VNIR + SWIR 9-band stack, and to the 11-band data set of SWIR and TIR data sets of the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor, using field spectra as training sets for evaluating the potential of these data sets in discriminating ophiolite rock units. Output results were compared with the geological map of the area and field observations, and were assessed by the use of confusion matrices. The assessment showed, in terms of kappa coefficient, that the SFF classification method with continuum removal applied to the SWIR data achieved excellent results, which were distinctively better than those obtained using VNIR + SWIR data and TIR data alone.     

Strategic effort allocation in online innovation tournaments

Online innovation tournaments, such as those hosted by crowdsourcing platforms (e.g., Kaggle), have been widely adopted by firms to evolve creative solutions to various problems. Solvers compete in these tournaments to earn rewards. In such competitive environments, it is imperative that solvers provide creative solutions with minimum effort. This article explores the factors that influence the solvers’ effort allocation decisions in a dynamic tournament setting. Specifically, comprehensive time variant data of teams that participated in crowdsourcing competitions on Kaggle were analyzed to gain insight into how solvers continually formulate strategies in light of performance feedback obtained through interim ranking. The results suggest that solvers strategically allocate their efforts throughout the contest to dynamically optimize their payoffs through balancing the probability of winning and the cost of expending effort. In particular, solvers tend to increase their efforts toward the end of tournaments or when they get closer to winning positions. Furthermore, our findings indicate that a last-minute surge in effort is more prevalent among high-skill solvers than in those with lower skill levels. In addition to providing insights that may help solvers develop strategies to improve their performance, the study has implications for the design of online crowdsourcing platforms, particularly in terms of incentivizing solvers to put forth their best effort.     

Fast Parallel Reordering and Isomorphism Testing of k -Trees

In this paper two problems on the class of k -trees, a subclass of the class of chordal graphs, are considered: the fast reordering problem and the isomorphism problem. An O(log ² n) time parallel algorithm for the fast reordering problem is described that uses O(nk(n-k)/\kern -1ptlog n) processors on a CRCW PRAM proving membership in the class NC for fixed k . An O(nk(k+1)!) time sequential algorithm for the isomorphism problem is obtained representing an improvement over the O(n ² k(k+1)!) algorithm of Sekharan (the second author) [10]. A parallel version of this sequential algorithm is presented that runs in O(log ² n) time using O((nk((k+1)!+n-k))/log n) processors improving on a parallel algorithm of Sekharan for the isomorphism problem [10]. Both the sequential and parallel algorithms use a concept introduced in this paper called the kernel of a k -tree.     

Rapid Concept Learning for Mobile Robots

Concept learning in robotics is an extremely challenging problem: sensory data is often high-dimensional, and noisy due to specularities and other irregularities. In this paper, we investigate two general strategies to speed up learning, based on spatial decomposition of the sensory representation, and simultaneous learning of multiple classes using a shared structure. We study two concept learning scenarios: a hallway navigation problem, where the robot has to induce features such as opening or wall. The second task is recycling, where the robot has to learn to recognize objects, such as a trash can. We use a common underlying function approximator in both studies in the form of a feedforward neural network, with several hundred input units and multiple output units. Despite the high degree of freedom afforded by such an approximator, we show the two strategies provide sufficient bias to achieve rapid learning. We provide detailed experimental studies on an actual mobile robot called PAVLOV to illustrate the effectiveness of this approach.