The quasi-Newton methods generally provide curvature information by approximating the Hessian using the secant equation. However, the secant equation becomes insipid in approximating the Newton step owing to its use of the first-order derivatives. In this study, we propose an approximate Newton step-based stochastic optimization algorithm for large-scale empirical risk minimization of convex functions with linear convergence rates. Specifically, we compute a partial column Hessian of size ($d\times k$) with $k\ll d$ randomly selected variables, then use the \textit{Nystr\"om method} to better approximate the full Hessian matrix. To further reduce the computational complexity per iteration, we directly compute the update step ($\Delta\boldsymbol{w}$) without computing and storing the full Hessian or its inverse. Furthermore, to address large-scale scenarios in which even computing a partial Hessian may require significant time, we used distribution-preserving (DP) sub-sampling to compute a partial Hessian. The DP sub-sampling generates $p$ sub-samples with similar first and second-order distribution statistics and selects a single sub-sample at each epoch in a round-robin manner to compute the partial Hessian. We integrate our approximated Hessian with stochastic gradient descent and stochastic variance-reduced gradients to solve the logistic regression problem. The numerical experiments show that the proposed approach was able to obtain a better approximation of Newton\textquotesingle s method with performance competitive with the state-of-the-art first-order and the stochastic quasi-Newton methods.
Biological data are generally high-dimensional and require efficient machine learning methods that are well generalized and scalable to discover their complex nonlinear patterns. The recent advances in the domain of artificial intelligence and machine learning can be attributed to deep neural networks (DNNs) because they accomplish a variety of tasks in computer vision and natural language processing. However, standard DNNs are not suitable for handling high-dimensional data and data with small number of samples because they require a large pool of computing resources as well as plenty of samples to learn a large number of parameters. In particular, although interpretability is important for high-dimensional biological data such as gene expression data, a nonlinear feature selection algorithm for DNN models has not been fully investigated. In this paper, we propose a novel nonlinear feature selection method called the Feature Selection Network (FsNet), which is a scalable concrete neural network architecture, under high-dimensional and small number of samples setups. Specifically, our network consists of a selector layer that uses a concrete random variable for discrete feature selection and a supervised deep neural network regularized with the reconstruction loss. Because a large number of parameters in the selector and reconstruction layer can easily cause overfitting under a limited number of samples, we use two tiny networks to predict the large virtual weight matrices of the selector and reconstruction layers. The experimental results on several real-world high-dimensional biological datasets demonstrate the efficacy of the proposed approach.
Graph structured data has wide applicability in various domains such as physics, chemistry, biology, computer vision, and social networks, to name a few. Recently, graph neural networks (GNN) were shown to be successful in effectively representing graph structured data because of their good performance and generalization ability. GNN is a deep learning based method that learns a node representation by combining specific nodes and the structural/topological information of a graph. However, like other deep models, explaining the effectiveness of GNN models is a challenging task because of the complex nonlinear transformations made over the iterations. In this paper, we propose GraphLIME, a local interpretable model explanation for graphs using the Hilbert-Schmidt Independence Criterion (HSIC) Lasso, which is a nonlinear feature selection method. GraphLIME is a generic GNN-model explanation framework that learns a nonlinear interpretable model locally in the subgraph of the node being explained. More specifically, to explain a node, we generate a nonlinear interpretable model from its $N$-hop neighborhood and then compute the K most representative features as the explanations of its prediction using HSIC Lasso. Through experiments on two real-world datasets, the explanations of GraphLIME are found to be of extraordinary degree and more descriptive in comparison to the existing explanation methods.