Abstract: This article introduces DCOT-LS-SVMs, a least-squares support vector machine-based deep cross-output knowledge transfer method. It improves the generalizability of least-squares support vector machines (LS-SVMs) without the complicated parameter tuning of kernel machines. The approach…
Abstract: Data mining and machine learning use matrix factorization (MF), an unsupervised learning technique for data representation. Different application scenarios can impose different constraints on the factorization to find the desired basis, which captures high-level…
Abstract: Averaging predictions from multiple models improves machine-learning algorithm performance. Multiple model performance optimization should generalize well. Models must share generalization knowledge. This article proposes a multi-kernel mutual learning method for hyperspectral classification using transfer…
Abstract: Classification is challenged by data collection frequency in some applications. We study mixed frequency data as a special type of multi-view data with different sampling frequencies for each view. DPLOL-MF is a discriminative dictionary…
Abstract: Unsupervised domain adaptation (UDA) uses a related but different distribution labeled source domain to learn a classifier for an unlabeled target domain. Most methods learn domain-invariant features by adapting image data. However, forcing domain-specific…
Abstract: Deep learning revived interest in neural architectures that can process complex graph-based structures, inspired by Graph Neural Networks (GNNs). The Scarselli et al. 2009 GNN model encodes the graph nodes’ states using an iterative…
Abstract: Database and AI are complementary. AI4DB makes databases smarter. Traditional empirical database optimization methods like cost estimation, join order selection, knob tuning, index and view selection cannot meet the high-performance requirements for large-scale database…
Abstract: Recommendation, CTR prediction, and image recognition use wide models like generalized linear models and factorization-based models. The memory-bounded models limit CPU performance improvement. GPUs have many computation units and high memory bandwidth, making them…
Abstract: Machine learning is increasingly using multiview learning. Most multiview learning methods cannot directly handle multiview sequential data, which often ignores dynamical structure. Most traditional multiview machine learning methods assume that sequence items at different…
Abstract: Pattern recognition, data mining, and machine learning all face clustering issues. Traditional shallow-structured clustering algorithms cannot uncover the interdependence of complex data features in latent space. Since they can learn promising latent representations from…
Abstract: Most scientific fields estimate mathematical model parameters. When processes and model descriptions become complex and there is no explicit likelihood function, this problem can be difficult. We propose BayesFlow, an invertible neural network-based globally…
Abstract: Many mobile social network users share messages. Spammers post links to viruses and ads or follow many users, resulting in many misleading messages in mobile social networks. This paper proposes an adaptive social spammer…
WhatsApp us