Maximizing the area under the receiver operating characteristic curve (AUC) is a standard approach to imbalanced classification. So far, various supervised AUC optimization methods have been developed and they are also extended to semi-supervised scenarios to cope with small sample problems. However, existing semi-supervised AUC optimization methods rely on strong distributional assumptions, which are rarely satisfied in real-world problems. In this paper, we propose a novel semi-supervised AUC optimization method that does not require such restrictive assumptions. We first develop an AUC optimization method based only on positive and unlabeled data (PU-AUC) and then extend it to semi-supervised learning by combining it with a supervised AUC optimization method. We theoretically prove that, without the restrictive distributional assumptions, unlabeled data contribute to improving the generalization performance in PU and semi-supervised AUC optimization methods. Finally, we demonstrate the practical usefulness of the proposed methods through experiments.

This paper presents a general graph representation learning framework called DeepGL for learning deep node and edge representations from large (attributed) graphs. In particular, DeepGL begins by deriving a set of base features (e.g., graphlet features) and automatically learns a multi-layered hierarchical graph representation where each successive layer leverages the output from the previous layer to learn features of a higher-order. Contrary to previous work, DeepGL learns relational functions (each representing a feature) that generalize across-networks and therefore useful for graph-based transfer learning tasks. Moreover, DeepGL naturally supports attributed graphs, learns interpretable features, and is space-efficient (by learning sparse feature vectors). In addition, DeepGL is expressive, flexible with many interchangeable components, efficient with a time complexity of $\mathcal{O}(|E|)$, and scalable for large networks via an efficient parallel implementation. Compared with the state-of-the-art method, DeepGL is (1) effective for across-network transfer learning tasks and attributed graph representation learning, (2) space-efficient requiring up to 6x less memory, (3) fast with up to 182x speedup in runtime performance, and (4) accurate with an average improvement of 20% or more on many learning tasks.

The area of Handwritten Signature Verification has been broadly researched in the last decades, but remains an open research problem. The objective of signature verification systems is to discriminate if a given signature is genuine (produced by the claimed individual), or a forgery (produced by an impostor). This has demonstrated to be a challenging task, in particular in the offline (static) scenario, that uses images of scanned signatures, where the dynamic information about the signing process is not available. Many advancements have been proposed in the literature in the last 5-10 years, most notably the application of Deep Learning methods to learn feature representations from signature images. In this paper, we present how the problem has been handled in the past few decades, analyze the recent advancements in the field, and the potential directions for future research.

We develop a method for incorporating relevant non-Euclidean geometric information into a broad range of classical filtering and statistical or machine learning algorithms. We apply these techniques to approximate the solution of the non-Euclidean filtering problem to arbitrary precision. We then extend the particle filtering algorithm to compute our asymptotic solution to arbitrary precision. Moreover, we find explicit error bounds measuring the discrepancy between our locally triangulated filter and the true theoretical non-Euclidean filter. Our methods are motivated by certain fundamental problems in mathematical finance. In particular we apply these filtering techniques to incorporate the non-Euclidean geometry present in stochastic volatility models and optimal Markowitz portfolios. We also extend Euclidean statistical or machine learning algorithms to non-Euclidean problems by using the local triangulation technique, which we show improves the accuracy of the original algorithm. We apply the local triangulation method to obtain improvements of the (sparse) principal component analysis and the principal geodesic analysis algorithms and show how these improved algorithms can be used to parsimoniously estimate the evolution of the shape of forward-rate curves. While focused on financial applications, the non-Euclidean geometric techniques presented in this paper can be employed to provide improvements to a range of other statistical or machine learning algorithms and may be useful in other areas of application.

In my last post, I covered the introduction to Regularization in supervised learning models. In this post, let's go over some of the regularization techniques widely used and the key difference between those. A regression model that uses L1 regularization technique is called Lasso Regression and model which uses L2 is called Ridge Regression. The key difference between these two is the penalty term. Ridge regression adds "squared magnitude" of coefficient as penalty term to the loss function.

This class is offered as CS7641 at Georgia Tech where it is a part of the Online Masters Degree (OMS). Taking this course here will not earn credit towards the OMS degree. The first part of the course covers Supervised Learning, a machine learning task that makes it possible for your phone to recognize your voice, your email to filter spam, and for computers to learn a bunch of other cool stuff. This class is offered as CS7641 at Georgia Tech where it is a part of the Online Masters Degree (OMS).

Any form of classification is supervised and not unsupervised[1][2]. You are probably interested in unsupervised image segmentation, where the algorithm attempts to determine which pixels are related and groups them into certain categories. This can be done by using traditional partitional clustering algorithms, such as K-means/EM[3], or advanced deep learning methods such as convolutional autoencoders[4], bayesian methods[5] and so on. You may read this survey research paper on the evaluation of such techniques - Image segmentation evaluation: A survey of unsupervised methods.

Policy evaluation or value function or Q-function approximation is a key procedure in reinforcement learning (RL). It is a necessary component of policy iteration and can be used for variance reduction in policy gradient methods. Therefore its quality has a significant impact on most RL algorithms. Motivated by manifold regularized learning, we propose a novel kernelized policy evaluation method that takes advantage of the intrinsic geometry of the state space learned from data, in order to achieve better sample efficiency and higher accuracy in Q-function approximation. Applying the proposed method in the Least-Squares Policy Iteration (LSPI) framework, we observe superior performance compared to widely used parametric basis functions on two standard benchmarks in terms of policy quality.

Detect facial keypoints is a critical element in face recognition. However, there is difficulty to catch keypoints on the face due to complex influences from original images, and there is no guidance to suitable algorithms. In this paper, we study different algorithms that can be applied to locate keyponits. Specifically: our framework (1)prepare the data for further investigation (2)Using PCA and LBP to process the data (3) Apply different algorithms to analysis data, including linear regression models, tree based model, neural network and convolutional neural network, etc. Finally we will give our conclusion and further research topic. A comprehensive set of experiments on dataset demonstrates the effectiveness of our framework.

We investigate a family of regression problems in a semi-supervised setting. The task is to assign real-valued labels to a set of $n$ sample points, provided a small training subset of $N$ labeled points. A goal of semi-supervised learning is to take advantage of the (geometric) structure provided by the large number of unlabeled data when assigning labels. We consider random geometric graphs, with connection radius $\epsilon(n)$, to represent the geometry of the data set. Functionals which model the task reward the regularity of the estimator function and impose or reward the agreement with the training data. Here we consider the discrete $p$-Laplacian regularization. We investigate asymptotic behavior when the number of unlabeled points increases, while the number of training points remains fixed. We uncover a delicate interplay between the regularizing nature of the functionals considered and the nonlocality inherent to the graph constructions. We rigorously obtain almost optimal ranges on the scaling of $\epsilon(n)$ for the asymptotic consistency to hold. We prove that the minimizers of the discrete functionals in random setting converge uniformly to the desired continuum limit. Furthermore we discover that for the standard model used there is a restrictive upper bound on how quickly $\epsilon(n)$ must converge to zero as $n \to \infty$. We introduce a new model which is as simple as the original model, but overcomes this restriction.