The paper deals with the application of Machine learning methods to the computerized recognition of separated handwritten or printed characters. The recognition system makes use of Instance-Based learning algorithms (...
详细信息
Cancer is the source of nearly one in six fatalities. There are numerous cancer types, breast cancer being one of them. IT can be easily detected by certain symptoms. Breast cancer affects not only women but also men;...
详细信息
Vision impairment has a profound influence on quality of life. This proposed work aims to work as an eye for the blind people. It will take in pictures as an input and interpret the details in the given picture as cap...
详细信息
This paper proposes a new source reconstruction method (SRM) based on deep learning. The conventional SRM usually requires oversampled measurements data to ensure higher accuracy. Thus, conventional SRM numerical syst...
详细信息
In view of fuzzy sets and their operations, three kinds of logic neurons, i.e., AND, OR and AND/OR neurons, are present in this paper. And those neurons can be classified into two types: weighted and relational. Using...
详细信息
ISBN:
(纸本)0780337964;0780337972
In view of fuzzy sets and their operations, three kinds of logic neurons, i.e., AND, OR and AND/OR neurons, are present in this paper. And those neurons can be classified into two types: weighted and relational. Using AND, OR and AND/OR neurons, a fuzzy relational model for dynamic system is provided as well as its learning algorithms. By a simple example, the soundness and the learning capability of the algorithms are verified.
Objects look very different in the underwater environment compared to their appearance in sunlight. High quality images with correct colouring simplify the detection of underwater objects. Hence, image processing is r...
详细信息
This paper compares the application of different neural models-multilayer perceptrons, radial basis functions and B-splines - for a benchmark problem, and illustrates the applicability of a common learning algorithm f...
详细信息
The ability to learn from past experiences and adapt one’s behavior accordingly within an environment or context to achieve a certain goal is a characteristic of a truly intelligent entity. Developing efficient, robu...
详细信息
The ability to learn from past experiences and adapt one’s behavior accordingly within an environment or context to achieve a certain goal is a characteristic of a truly intelligent entity. Developing efficient, robust, and reliable learning algorithms towards that end is an active area of research and a major step towards achieving artificial general intelligence. In this thesis, we research learning algorithms for optimal decision making in two different contexts, Reinforcement learning in Part I and Auction Design in Part II. Reinforcement learning (RL) is an area of machine learning that is concerned with how an agent should act in an environment in order to maximize its cumulative reward over time. In Chapter 2, inspired by statistical physics, we develop a novel approach to RL that not only learns optimal policies with enhanced desirable properties but also sheds new light on maximum entropy RL. In Chapter 3, we tackle the generalization problem in RL using a Bayesian perspective. We show that imperfect knowledge of the environment’s dynamics effectively turn a fully-observed Markov Decision Process (MDP) into a Partially Observed MDP (POMDP) that we call the Epistemic POMDP. Informed by this observation, we develop a new policy learning algorithm LEEP which has improved generalization properties. An auction is the process of organizing the buying and selling of products and services that is of great practical importance. Designing an incentive compatible, individually rational auction that maximizes revenue is a challenging and intractable problem. Recently, a deep learning based approach was proposed to learn optimal auctions from data. While successful, this approach suffers from a few limitations, including sample inefficiency, lack of generalization to new auctions, and training difficulties. In Chapter 4, we construct a symmetry preserving neural network architecture, EquivariantNet, suitable for anonymous auctions. EquivariantNet is not only more sample
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 method for constru...
详细信息
The use of multiple antennas in mobile devices provides enhanced data rates at the cost of increased power consumption. The stochastic nature of the wireless propagation medium and random variations in the utilization...
详细信息
ISBN:
(纸本)9781467310680
The use of multiple antennas in mobile devices provides enhanced data rates at the cost of increased power consumption. The stochastic nature of the wireless propagation medium and random variations in the utilization and operating environment of the device makes it difficult to estimate and predict wireless channels and power consumption levels. Therefore, we investigate a robust antenna subset selection policy where the power-normalized throughput is assumed to be drawn from an unknown distribution with unknown mean. At each time instant, the transceiver decides upon the active antenna subset based on observations of the outcomes of previous choices, with the objective being to identify the optimal antenna subset which maximizes the power-normalized throughput. In this work, we present a sequential learning scheme to achieve this based on the theory of multi-armed bandits. Simulations verify that the proposed novel method that accounts for dependent arms outperforms a naive approach designed for independent arms in terms of regret.
暂无评论