We introduce a model that learns active learning algorithms via metalearning. For a distribution of related tasks, our model jointly learns: a data representation, an item selection heuristic, and a prediction functio...
详细信息
ISBN:
(纸本)9781510855144
We introduce a model that learns active learning algorithms via metalearning. For a distribution of related tasks, our model jointly learns: a data representation, an item selection heuristic, and a prediction function. Our model uses the item selection heuristic to construct a labeled support set for training the prediction function. Using the Omniglot and MovieLens datasets, we test our model in synthetic and practical settings.
In this paper, we consider the link prediction problem, where we are given a partial snapshot of a network at some time and the goal is to predict the additional links formed at a later time. The accuracy of current p...
详细信息
ISBN:
(纸本)9783642158797
In this paper, we consider the link prediction problem, where we are given a partial snapshot of a network at some time and the goal is to predict the additional links formed at a later time. The accuracy of current prediction methods is quite low due to the extreme class skew and the large number of potential links. Here, we describe learning algorithms based on chance constrained programs and show that they exhibit all the properties needed for a good link predictor, namely, they allow preferential bias to positive or negative class;handle skewness in the data;and scale to large networks. Our experimental results on three real-world domains-co-authorship networks, biological networks and citation networks-show significant performance improvement over baseline algorithms. We conclude by briefly describing some promising future directions based on this work.
A relatively new approach to maximizing the probability of target detection in airborne antenna radar is to implement a linear filter called an adaptive space-time processor (STP). Theoretically, a fully-adaptive STP ...
详细信息
ISBN:
(纸本)0780337263
A relatively new approach to maximizing the probability of target detection in airborne antenna radar is to implement a linear filter called an adaptive space-time processor (STP). Theoretically, a fully-adaptive STP can implement the optimal solution (a Wiener filter) in real-time. In practice, though, fully-adaptive STPs present several problems. First, the number of weights in the linear filter can be extremely large. Second, the computational complexity a single weight vector is on the order of O(MN)(3). Third, the STP is a linear filter but the operating environment is nonlinear and nonstationary. The consequences arising from these problems is that real-time, optimal processing using the traditional techniques of space-time processors is beyond current computing technology. In this paper we explore the applicability of artificial neural networks and learning algorithms for minimizing the effect of motion clutter on target detection. Artificial neural networks are adaptive, parallel, distributed processing systems capable of performing complex computations in real time. learning algorithms are the mechanisms by which the long-term memory in artificial neural networks is updated, but not destroyed, to accommodate new or changing information. Because learning algorithms retain information over the lifetime of the system, but are also modifiable, they can minimizing the computational requirements faced by radar systems, yet still adapt to changing environmental conditions.
The quantron is a new artificial neuron model, able to solve nonlinear classification problems, for which an efficient learning algorithm has yet to be developed. Using surrogate potentials, constraints on some parame...
详细信息
ISBN:
(纸本)9781424496365
The quantron is a new artificial neuron model, able to solve nonlinear classification problems, for which an efficient learning algorithm has yet to be developed. Using surrogate potentials, constraints on some parameters and an infinite number of potentials, we obtain analytical expressions involving ceiling functions for the activation function of the quantron. We then show how to retrieve the parameters of a neuron from the images it produced.
Collinear (flat) pattern appears in a given set of multidimensional feature vectors when many of these vectors are located on (or near) some plane in the feature space. Flat pattern discovered in a given data set can ...
详细信息
ISBN:
(纸本)9783662493908;9783662493892
Collinear (flat) pattern appears in a given set of multidimensional feature vectors when many of these vectors are located on (or near) some plane in the feature space. Flat pattern discovered in a given data set can give indications for creating a model of interaction between selected features. Patterns located on planes can be discovered even in large and multidimensional data sets through minimization of the convex and piecewise linear (CPL) criterion functions. Discovering flat patterns can be based on the search for degenerated vertices in the parameter space. The possibility of using learning algorithms for this purpose is examined in this paper.
Dynamic hedging is a financial strategy that consists in periodically transacting one or multiple financial assets to offset the risk associated with a correlated liability. Deep Reinforcement learning (DRL) algorithm...
详细信息
Although gossip and random walk-based learning algorithms are widely known for decentralized learning, there has been limited theoretical and experimental analysis to understand their relative performance for differen...
详细信息
The aim of the paper is to present a conception of learning algorithms for discrete manufacturing processes control. A general knowledge based model of a vast class of discrete manufacturing processes (DMP) is given. ...
详细信息
ISBN:
(纸本)3540357483
The aim of the paper is to present a conception of learning algorithms for discrete manufacturing processes control. A general knowledge based model of a vast class of discrete manufacturing processes (DMP) is given. The model is a basis for the method of the synthesis of intelligent, learning algorithms that use information on the process gained in previous iterations as well as an expert knowledge. To illustrate the presented ideas, the scheduling algorithm for a special NP-hard problem is given.
This tutorial is on applications of computational learning theory to verification of systems. Computational learning theory deals with algorithmic models for learning formally representable concepts using either posit...
详细信息
暂无评论