Deep Gaussian Processes (DGPs) are hierarchical generalizations of Gaussian Processes that combine well calibrated uncertainty estimates with the high flexibility of multilayer models. One of the biggest challenges with these models is that exact inference is intractable. The current state-of-the-art inference method, Variational Inference (VI), employs a Gaussian approximation to the posterior distribution. This can be a potentially poor unimodal approximation of the generally multimodal posterior. In this work, we provide evidence for the non-Gaussian nature of the posterior and we apply the Stochastic Gradient Hamiltonian Monte Carlo method to directly sample from it. To efficiently optimize the hyperparameters, we introduce the Moving Window MCEM algorithm. This results in significantly better predictions at a lower computational cost than its VI counterpart. Thus our method establishes a new state-of-the-art for inference in DGPs.

Take my free 7-day email crash course now (with sample code). Click to sign-up and also get a free PDF Ebook version of the course.

We address the problem of predicting the labeling of a graph in an online setting when the labeling is changing over time. We provide three mistake-bounded algorithms based on three paradigmatic methods for online algorithm design. The algorithm with the strongest guarantee is a quasi-Bayesian classifier which requires $\mathcal{O}(t \log n)$ time to predict at trial $t$ on an $n$-vertex graph. The fastest algorithm (with the weakest guarantee) is based on a specialist [10] approach and surprisingly only requires $\mathcal{O}(\log n)$ time on any trial $t$. We also give an algorithm based on a kernelized Perceptron with an intermediate per-trial time complexity of $\mathcal{O}(n)$ and a mistake bound which is not strictly comparable. Finally, we provide experiments on simulated data comparing these methods.

We consider the problem of learning to play first-person shooter (FPS) video games using raw screen images as observations and keyboard inputs as actions. The high-dimensionality of the observations in this type of applications leads to prohibitive needs of training data for model-free methods, such as the deep Q-network (DQN), and its recurrent variant DRQN. Thus, recent works focused on learning low-dimensional representations that may reduce the need for data. This paper presents a new and efficient method for learning such representations. Salient segments of consecutive frames are detected from their optical flow, and clustered based on their feature descriptors. The clusters typically correspond to different discovered categories of objects. Segments detected in new frames are then classified based on their nearest clusters. Because only a few categories are relevant to a given task, the importance of a category is defined as the correlation between its occurrence and the agent's performance. The result is encoded as a vector indicating objects that are in the frame and their locations, and used as a side input to DRQN. Experiments on the game Doom provide a good evidence for the benefit of this approach.

Due to the rapid innovation of technology and the desire to find and employ biomarkers for neurodegenerative disease, high-dimensional data classification problems are routinely encountered in neuroimaging studies. To avoid over-fitting and to explore relationships between disease and potential biomarkers, feature learning and selection plays an important role in classifier construction and is an important area in machine learning. In this article, we review several important feature learning and selection techniques including lasso-based methods, PCA, the two-sample t-test, and stacked auto-encoders. We compare these approaches using a numerical study involving the prediction of Alzheimer's disease from Magnetic Resonance Imaging.

We propose a very simple modification of gradient descent and stochastic gradient descent. We show that when applied to a variety of machine learning models including softmax regression, convolutional neural nets, generative adversarial nets, and deep reinforcement learning, this very simple surrogate can dramatically reduce the variance and improve the accuracy of the generalization. The new algorithm, (which depends on one nonnegative parameter) when applied to non-convex minimization, tends to avoid sharp local minima. Instead it seeks somewhat flatter local (and often global) minima. The method only involves preconditioning the gradient by the inverse of a tri-diagonal matrix that is positive definite. The motivation comes from the theory of Hamilton-Jacobi partial differential equations. This theory demonstrates that the new algorithm is almost the same as doing gradient descent on a new function which (a) has the same global minima as the original function and (b) is "more convex". Again, the programming effort in doing this is minimal, in cost, complexity and effort. We implement our algorithm into both PyTorch and Tensorflow platforms, which will be made publicly available.

Appropriate treatment regimens play a vital role in improving patient health status. Although some achievements have been made, few of the recent studies of learning treatment regimens have exploited different kinds of patient information due to the difficulty in adopting heterogeneous data to many data mining methods. Moreover, current studies seem too rigid with fixed intervals of treatment periods corresponding to the varying lengths of hospital stay. To this end, this work proposes a generic data-driven framework which can derive group-treatment regimens from electronic medical records by utilizing a mixed-variate restricted Boltzmann machine and incorporating medical domain knowledge. We conducted experiments on coronary artery disease as a case study. The obtained results show that the framework is promising and capable of assisting physicians in making clinical decisions.

There is no free lunch, no single learning algorithm that will outperform other algorithms on all data. In practice different approaches are tried and the best algorithm selected. An alternative solution is to build new algorithms on demand by creating a framework that accommodates many algorithms. The best combination of parameters and procedures is searched here in the space of all possible models belonging to the framework of Similarity-Based Methods (SBMs). Such meta-learning approach gives a chance to find the best method in all cases. Issues related to the meta-learning and first tests of this approach are presented.

Machine Learning Key Terms Posted on Monday, June 4th, 2018 at 3:03 pm. Share this: This blog post was authored by Soniya Shah. Machine learning seems to be everywhere these days – in the online recommendations you get on Netflix, the self-driving cars that hyped in the media, and in serious cases, like fraud detection. Data is a huge part of machine learning, and so are the key terms. Unless you have a background in statistics or data science, it can be confusing to keep all the terminology straight. And even then, you might want to keep a list of terms handy.

This post is an experiment combining the result of t-SNE with two well known clustering techniques: k-means and hierarchical. This will be the practical section, in R. But also, this post will explore the intersection point of concepts like dimension reduction, clustering analysis, data preparation, PCA, HDBSCAN, k-NN, SOM, deep learning...and Carl Sagan! For those who don't know t-SNE technique (official site), it's a projection technique -or dimension reduction- similar in some aspects to Principal Component Analysis (PCA), used to visualize N variables into 2 (for example). When the t-SNE output is poor Laurens van der Maaten (t-SNE's author) says: As a sanity check, try running PCA on your data to reduce it to two dimensions.