We connect learning algorithms and algorithms automating proof search in propositional proof systems: for every sufficiently strong, well-behaved propositional proof system P, we prove that the following statements ar...
详细信息
In a principal-agent problem, a principal seeks to motivate an agent to take a certain action beneficial to the principal, while spending as little as possible on the reward. This is complicated by the fact that the p...
详细信息
ISBN:
(纸本)1595933832
In a principal-agent problem, a principal seeks to motivate an agent to take a certain action beneficial to the principal, while spending as little as possible on the reward. This is complicated by the fact that the principal does not know the agent's utility function (or type). We study the online setting where at each round, the principal encounters a new agent, and the principal sets the rewards anew. At the end of each round, the principal only finds out the action that the agent took, but not his type. The principal must learn how to set the rewards optimally. We show that this setting generalizes the setting of selling a digital good online. We study and experimentally compare three main approaches to this problem. First, we show how to apply a standard bandit algorithm to this setting. Second, for the case where the distribution of agent types is fixed (but unknown to the principal), we introduce a new gradient ascent algorithm. Third, for the case where the distribution of agents' types is fixed, and the principal has a prior belief (distribution) over a limited class of type distributions, we study a Bayesian approach.
The productivity of milling processes is limited by the occurrence of chatter vibrations. The correlation of the maximum stable cutting depth and the spindle speed can be shown in a stability lobe diagram (SLD). The s...
详细信息
The productivity of milling processes is limited by the occurrence of chatter vibrations. The correlation of the maximum stable cutting depth and the spindle speed can be shown in a stability lobe diagram (SLD). The stability is different for different width of cut and can change with the axis positions. Today it is already a great effort to estimate the SLD only for one position. Many experiments are necessary to measure the SLD or derive a detailed mathematical model to calculate the SLD. Moreover not only the cutting depth, but also the cutting width should be represented in the SLD. This paper presents a new approach to assess the process stability based on measured acceleration signals. The multidimensional stability lobe diagram (MSLD) are derived during the production using two new continuously learning algorithms. In this paper the application of a continuous learning support vector machine and a continuous neural network is shown. The support vector machine and the neural network are extended to make them capable for continuous learning and time-variant systems. A new trust criterion is introduced, which gives information about the prediction quality of the output for the selected input region. The learned MSLDs are evaluated against analytically calculated MSLDs"and the learning algorithms can reproduce the analytical results very well. (C) 2015 Elsevier Ltd. All rights reserved.
This paper proposes supervised learning algorithms based on gradient descent for training reformulated radial basis function (RBF) neural networks. Such RBF models employ radial basis functions whose form is determine...
详细信息
This paper proposes supervised learning algorithms based on gradient descent for training reformulated radial basis function (RBF) neural networks. Such RBF models employ radial basis functions whose form is determined by admissible generator functions. RBF networks with Gaussian radial basis functions are generated by exponential generator functions. A sensitivity analysis provides the basis for selecting generator functions by investigating the effect of linear, exponential and logarithmic generator functions on gradient descent learning. Experiments involving reformulated RBF networks indicate that the proposed gradient descent algorithms guarantee fast learning and very satisfactory function approximation capability.
The use of learning algorithms to automatically set up analytical procedures in the near IR leads to considerable statistical complexities when one attempts to refine and optimize the method. While current work employ...
详细信息
The synchronization of energy supply and demand will pose major challenges for the future energy market. A necessary requirement for this is reliable energy consumption forecasts. learning algorithms have great potent...
详细信息
We present a model that learns active learning algorithms via metalearning. For each metatask, our model jointly learns: a data representation, an item selection heuristic, and a one-shot classifier. Our model uses th...
详细信息
In this paper, two types of the learning algorithms of the GMDH (Group Method of Data Handling) neural network are proposed. By using the heuristic self-organization method, the GMDH neural network can automatically o...
详细信息
In this paper, two types of the learning algorithms of the GMDH (Group Method of Data Handling) neural network are proposed. By using the heuristic self-organization method, the GMDH neural network can automatically organized the optimal neural network structure which fit for the complexity of the nonlinear system. In the learning calculations, the neural network structure which is organized by the GMDH neural network is not changed and the values of the weights between the neurons are renewed. The GMDH neural network algorithm and its learning algorithms are applied to the medical image recognition.
A fuzzy model has the ability to represent relationships that are too complex or not well enough understood to be directly described by precise mathematical models. In domains with one and two input dimensions, the co...
详细信息
A fuzzy model has the ability to represent relationships that are too complex or not well enough understood to be directly described by precise mathematical models. In domains with one and two input dimensions, the combined learning algorithms and completion produce models with a specified degree of accuracy with considerably fewer training examples than those using the learning algorithm without completion. The effectiveness of the combination of learning and completion in problem domains of higher dimensions is examined. The components of the hierarchical model are reviewed. The results of the experiments designed to illustrate the effects of dimensionality on the learning algorithms are discussed.
Three deep learning based algorithms are compared in terms of standard figures of merit and visual quality for single image deraining. The comparative analysis additionally considers several classic approaches based o...
详细信息
暂无评论