Pairwise learning usually refers to a learning task which involves a loss function depending on pairs of examples, among which most notable ones are bipartite ranking, metric learning and AUC maximization. In this pap...
详细信息
Pairwise learning usually refers to a learning task which involves a loss function depending on pairs of examples, among which most notable ones are bipartite ranking, metric learning and AUC maximization. In this paper, we focus on online learning algorithms for pairwise learning problems without strong convexity, for which all previously known algorithms achieve a convergence rate of O(1/root T) after T iterations. In particular, we study an online learning algorithm for pairwise learning with a least-square loss function in an unconstrained setting. We prove that the convergence of its last iterate can converge to the desired minimizer at a rate arbitrarily close to O(1/T) up to logarithmic factor. The rates for this algorithm are established in high probability under the assumptions of polynomially decaying step sizes.
In recent years a large number of digital mammograms have been generated in hospitals and breast screening centers. To assist diagnosis through indexing those mammogram databases, we proposed a content-based image ret...
详细信息
ISBN:
(纸本)3540356258
In recent years a large number of digital mammograms have been generated in hospitals and breast screening centers. To assist diagnosis through indexing those mammogram databases, we proposed a content-based image retrieval framework along with a novel feature extraction technique for describing the degree of calcification phenomenon revealed in the mammograms and six relevance feedback learning algorithms, which fall in the category of query point movement, for improving system performance. The results show that the proposed system can reach a precision rate of 0.716 after five rounds of relevance feedback have been performed.
This paper describes several new online model-free reinforcement learning (RL) algorithms. We designed three new reinforcement algorithms, namely: QV2, QVMAX, and QV-MAX2, that are all based on the QV-learning algorit...
详细信息
ISBN:
(纸本)9781424427611
This paper describes several new online model-free reinforcement learning (RL) algorithms. We designed three new reinforcement algorithms, namely: QV2, QVMAX, and QV-MAX2, that are all based on the QV-learning algorithm, but in contrary to QV-learning, QVMAX and QVMAX2 are off-policy RL algorithms and QV2 is a new on-policy RL algorithm. We experimentally compare these algorithms to a large number of different RL algorithms, namely: Q-learning, Sarsa, R-learning, Actor-Critic, QV-learning, and ACLA. We show experiments on five maze problems of varying complexity. Furthermore, we show experimental results on the cart pole balancing problem. The results show that for different problems, there can be large performance differences between the different algorithms, and that there is not a single RL algorithm that always performs best, although on average QV-learning scores highest.
In this paper, we consider a new scenario for privacy-preserving data mining called two-part partitioned record model (TPR) and find solutions for a family of frequency-based learning algorithms in TPR model. In TPR, ...
详细信息
ISBN:
(纸本)9783319028217;9783319028200
In this paper, we consider a new scenario for privacy-preserving data mining called two-part partitioned record model (TPR) and find solutions for a family of frequency-based learning algorithms in TPR model. In TPR, the dataset is distributed across a large number of users in which each record is owned by two different users, one user only knows the values for a subset of attributes and the other knows the values for the remaining attributes. A miner aims to learn, for example, classification rules on their data, while preserving each user's privacy. In this work we develop a cryptographic solution for frequency-based learning methods in TPR. The crucial step in the proposed solution is the privacy-preserving computation of frequencies of a tuple of values in the users' data, which can ensure each user's privacy without loss of accuracy. We illustrate the applicability of the method by using it to build the privacy preserving protocol for the naive Bayes classifier learning, and briefly address the solution in other applications. Experimental results show that our protocol is efficient.
This paper presents both a theoretical discussion and an experimental comparison of batch and incremental learning in an attempt to individuate some of the respective advantages and disadvantages of the two approaches...
详细信息
ISBN:
(纸本)3540650687
This paper presents both a theoretical discussion and an experimental comparison of batch and incremental learning in an attempt to individuate some of the respective advantages and disadvantages of the two approaches when learning from frequently updated databases. The paper claims that incremental learning might be more suitable for this purpose, although a number of issues remain to be resolved.
Incremental learning (IL) plays a key role in many real-world applications where data arrives over time. It is mainly concerned with learning models in an ever-changing environment. In this paper, we review some of th...
详细信息
ISBN:
(纸本)9781424412099
Incremental learning (IL) plays a key role in many real-world applications where data arrives over time. It is mainly concerned with learning models in an ever-changing environment. In this paper, we review some of the incremental learning algorithms and evaluate them within the same experimental settings in order to provide as objective comparative study as possible. These algorithms include fuzzy ARTMAP, nearest generalized exemplar, growing neural gas, generalized fuzzy min-max neural network, and IL based on function decomposition (ILFD).
We survey the fastest known algorithms for learning various expressive classes of Boolean functions in the Probably Approximately Correct (PAC) learning model.
ISBN:
(纸本)3540340211
We survey the fastest known algorithms for learning various expressive classes of Boolean functions in the Probably Approximately Correct (PAC) learning model.
In this paper we study Temporal Difference (TD) learning with linear value function approximation. The classic TD algorithm is known to be unstable with linear function approximation and off-policy learning. Recently ...
详细信息
ISBN:
(纸本)9781479945528
In this paper we study Temporal Difference (TD) learning with linear value function approximation. The classic TD algorithm is known to be unstable with linear function approximation and off-policy learning. Recently developed Gradient TD (GTD) algorithms have addressed this problem successfully. Despite their prominent properties of good scalability and convergence to correct solutions, they inherit the potential weakness of slow convergence as they are a stochastic gradient descent algorithm. Accelerated stochastic gradient descent algorithms have been developed to speed up convergence, while still keeping computational complexity low. In this work, we develop an accelerated stochastic gradient descent method for minimizing the Mean Squared Projected Bellman Error (MSPBE), and derive a bound for the Lipschitz constant of the gradient of the MSPBE, which plays a critical role in our proposed accelerated GTD algorithms. Our comprehensive numerical experiments demonstrate promising performance in solving the policy evaluation problem, in comparison to the GTD algorithm family. In particular, accelerated TDC surpasses state-of-the-art algorithms.
We provide new results for noise-tolerant and sample-efficient learning algorithms under s-concave distributions. The new class of s-concave distributions is a broad and natural generalization of log-concavity, and in...
详细信息
We provide new results for noise-tolerant and sample-efficient learning algorithms under s-concave distributions. The new class of s-concave distributions is a broad and natural generalization of log-concavity, and includes many important additional distributions, e.g., the Pareto distribution and t-distribution. This class has been studied in the context of efficient sampling, integration, and optimization, but much remains unknown about the geometry of this class of distributions and their applications in the context of learning. The challenge is that unlike the commonly used distributions in learning (uniform or more generally log-concave distributions), this broader class is not closed under the marginalization operator and many such distributions are fat-tailed. In this work, we introduce new convex geometry tools to study the properties of s-concave distributions and use these properties to provide bounds on quantities of interest to learning including the probability of disagreement between two halfspaces, disagreement outside a band, and the disagreement coefficient. We use these results to significantly generalize prior results for margin-based active learning, disagreement-based active learning, and passive learning of intersections of halfspaces. Our analysis of geometric properties of s-concave distributions might be of independent interest to optimization more broadly.
We develop an asymptotic framework to compare the test performance of (personalized) federated learning algorithms whose purpose is to move beyond algorithmic convergence arguments. To that end, we study a high-dimens...
详细信息
We develop an asymptotic framework to compare the test performance of (personalized) federated learning algorithms whose purpose is to move beyond algorithmic convergence arguments. To that end, we study a high-dimensional linear regression model to elucidate the statistical properties (per client test error) of loss minimizers. Our techniques and model allow precise predictions about the benefits of personalization and information sharing in federated scenarios, including that Federated Averaging with simple client fine-tuning achieves identical asymptotic risk to more intricate meta-learning approaches and outperforms naive Federated Averaging. We evaluate and corroborate these theoretical predictions on federated versions of the EMNIST, CIFAR-100, Shakespeare, and Stack Overflow datasets.
暂无评论