We establish learning rates to the Bayes risk for support vector machines (SVMs) with hinge loss. In particular, for SVMs with Gaussian RBF kernels we propose a geometric condition for distributions which can be used ...
ISBN:
(纸本)0262195348
We establish learning rates to the Bayes risk for support vector machines (SVMs) with hinge loss. In particular, for SVMs with Gaussian RBF kernels we propose a geometric condition for distributions which can be used to determine approximation properties of these kernels. Finally, we compare our methods with a recent paper of G. Blanchard et al.
Sparse representation-based classification(SRC) has been widely used because it just relies on simple linear regression ideas to do classification, and it does not need learning. However, the performance of SRC is lim...
详细信息
Sparse representation-based classification(SRC) has been widely used because it just relies on simple linear regression ideas to do classification, and it does not need learning. However, the performance of SRC is limited by needing sufficient labeled samples per class and the sensitivity to class imbalance. For tackling these problems, an improved SRC model is constructed in this paper. For alleviating the problem of insufficient labeled samples, an unlabeled data-driven inverse projection sparse representation-based classification model is constructed to achieve effective and stable representation and recognition results. The L1/2 and S1/2 regularizations are introduced to capture the sparsity of 1-D and 2-D, and to make the model have good statistical properties. Moreover, the cost-sensitive strategy is integrated into the model's classification criteria to improve the imbalance of class distribution adaptively, especially for multiclass imbalanced data.A solver utilizing the mixed Gauss-Seidel and Jacobian ADMM algorithm is developed to obtain the approximate solution. Experiments on common public test databases show that the proposed model achieves competitive results compared with the latest published results and some deep-learning algorithms.
Evolutionary algorithms are largely used search and optimization procedures that, when properly designed, can solve intractable problems in tractable polynomial time. Efficiency enhancements are used to turn them from...
详细信息
ISBN:
(纸本)9781605581309
Evolutionary algorithms are largely used search and optimization procedures that, when properly designed, can solve intractable problems in tractable polynomial time. Efficiency enhancements are used to turn them from tractable to practical. In this paper we show preliminary results of two efficiency enhancements proposed for the Extended Compact Genetic Algorithm. First, a model building enhancement was used to reduce the complexity of the process from O(n3) to O(n2), speeding up the algorithm by 1000 times on a 4096 bits problem. Then, local-search hybridization was used to reduce the population size by at least 32 times, reducing the memory and running time required by the algorithm. These results draw the first steps toward a competent and efficient Genetic Algorithm.
Genetic Programming (GP) is a dynamic field of research where empirical testing plays an important role in validating new ideas and algorithms. The ability to easily prototype new algorithms by reusing key components ...
详细信息
One fundamental problem that arises In VLSI layout analysis and verification Is the segment Intersection problem: given a set of segments In the plane, find all pairwise Intersections. This problem has been widely stu...
详细信息
ISBN:
(纸本)0780391977
One fundamental problem that arises In VLSI layout analysis and verification Is the segment Intersection problem: given a set of segments In the plane, find all pairwise Intersections. This problem has been widely studied In the Computational Geometry. One problem with processing large VLSI layouts Is that the data to be processed may be far too massive to fit In main memory. When dealing with data sets of sizes exceeding main memory, communication between the fast Internal memory and the slow external memory Is often the performance bottleneck and algorithms and data structures designed under the assumption of a single level of memory may be meaningless. External-memory algorithms try to optimize performance by taking Into account disk accesses. One can certainly use the standard main memory algorithms for data that reside on disk but their performance Is often considerably below the optimum because there Is no control over how the operating system performs disk accesses. On demand thrashing can be high thus resulting In an Increase In response time. Although a lot of research has been done In the recent past on efficient external-memory algorithms and data structures, such work In the area of VLSI computer-aided design Is limited. We have designed and Implemented a practical external-memory algorithm for reporting all Intersecting pairs amongst a set of orthogonal segments. The key to our success Is that we take advantage of the fact that real data sets from VLSI applications tend to obey the so-called "square-root" rule, I.e. In a set of N line segments, the expected number of line segments Intersecting a horizontal or vertical scanline In a VLSI layout Is O(√N), a fact Ignored by known external-memory algorithms. Another factor that Is crucial to our success Is that other algorithms stores the data structures In external memory requiring I/O to access them. We reduce such disk accesses by using a clever storage scheme. Our algorithm outperforms not only a standa
An intermediate level between neural circuits and behaviors is neural computations, various behaviors that animals exhibit following some basic control laws can be implemented by some canonical neural computations [Ca...
详细信息
We present a new construction of a compression function that uses two parallel calls to an ideal primitive (an ideal blockcipher or a public random function) from to bits. This is similar to the well-known MDC-2 or th...
详细信息
We show that anomaly detection can be interpreted as a binary classification problem. Using this interpretation we propose a support vector machine (SVM) for anomaly detection. We then present some theoretical results...
详细信息
ISBN:
(纸本)0262195348
We show that anomaly detection can be interpreted as a binary classification problem. Using this interpretation we propose a support vector machine (SVM) for anomaly detection. We then present some theoretical results which include consistency and learning rates. Finally, we experimentally compare our SVM with the standard one-class SVM.
In a range-aggegate query problem we wish to preprocess a set S of geometric objects such that given a query orthogonal range q, a certain intersection or proximity query on the objects of S intersected by q can be an...
详细信息
暂无评论