Regularized nonlinear acceleration (RNA) is a generic extrapolation scheme for optimization methods, with marginal computational overhead. It aims to improve convergence using only the iterates of simple iterative algorithms. However, so far its application to optimization was theoretically limited to gradient descent and other single-step algorithms. Here, we adapt RNA to a much broader setting including stochastic gradient with momentum and Nesterov's fast gradient. We use it to train deep neural networks, and empirically observe that extrapolated networks are more accurate, especially in the early iterations. A straightforward application of our algorithm when training ResNet-152 on ImageNet produces a top-1 test error of 20.88%, improving by 0.8% the reference classification pipeline. Furthermore, the code runs offline in this case, so it never negatively affects performance.

Many tasks in machine learning and signal processing can be solved by minimizing a convex function of a measure. This includes sparse spikes deconvolution or training a neural network with a single hidden layer. For these problems, we study a simple minimization method: the unknown measure is discretized into a mixture of particles and a continuous-time gradient descent is performed on their weights and positions. This is an idealization of the usual way to train neural networks with a large hidden layer. We show that, when initialized correctly and in the many-particle limit, this gradient flow, although non-convex, converges to global minimizers. The proof involves Wasserstein gradient flows, a by-product of optimal transport theory. Numerical experiments show that this asymptotic behavior is already at play for a reasonable number of particles, even in high dimension.

We design a new myopic strategy for a wide class of sequential design of experiment (DOE) problems, where the goal is to collect data in order to to fulfil a certain problem specific goal. Our approach, Myopic Posterior Sampling (MPS), is inspired by the classical posterior (Thompson) sampling algorithm for multi-armed bandits and leverages the flexibility of probabilistic programming and approximate Bayesian inference to address a broad set of problems. Empirically, this general-purpose strategy is competitive with more specialised methods in a wide array of DOE tasks, and more importantly, enables addressing complex DOE goals where no existing method seems applicable. On the theoretical side, we leverage ideas from adaptive submodularity and reinforcement learning to derive conditions under which MPS achieves sublinear regret against natural benchmark policies.

So far, we have investigated if Father Age and Mother Age were impacting Gestation Week, and we know that both Father Age and Mother Age influence the changes in Gestation Week. But since we have done the investigation separately, one for Father Age's influence on Gestation Week and another for Mother's Age's influence on Gestation Week, we still don't know which of Father Age and Mother Age is the direct cause of the influence. In this post, I'm going to investigate further to find this out. So far, we know that the increases in Father Age would make Gestation Week shorter. And, the increases in Mother Age would also make Gestation Week shorter.

Linear Regression is an algorithm that helps us predict unknown numeric outcome in future. It is usually the first Machine Learning (or Statistical) algorithms to learn when you are stepping into the world of Data Science or Machine Learning. Though it is one of the'old school' Statistical algorithms, it is still the most often used algorithm among many data scientists even today thanks to its simplicity and explainability. We at Exploratory always focus on, as the name suggests, making Exploratory Data Analysis (EDA) easier. EDA is a practice of iteratively asking a series of questions about data and trying to gain useful insights out of the data to answer the questions and essentially to influence our decision making.

With the development of wearable cameras, first-person activity recognition has been a popular topic in recent years [1]. There are many conventional approaches which tackle first-person activity recognition. Some of these approaches employ motion feature such as optical flow and also a classifier, e.g., LogitBoost and SVM (support vector machine)[2, 3]. In recent years, DCNN (deep convolutional neural network), the state-of-the-art model for visual recognition, has been proposed [4] and then applied to several tasks on first-person activity recognition. Although DCNN models provide remarkable results for image recognition, they require a large amount of labeled training samples.

As the amount of data continues to grow at an almost incomprehensible rate, being able to understand and process data is becoming a key differentiator for IT professionals and data-scientists. The scikit-learn library is one of the most popular platforms for everyday Machine Learning and data science because it is built upon Python, a fully featured programming language. This comprehensive 3-in-1 course is your one-stop solution to everything that matters in mastering machine learning algorithms and their implementation. Develop pipelines and process data through manipulation, extraction, and data-cleansing techniques. Learn clean coding techniques which are applicable to any scalable Machine Learning projects.

This course is a lead-in to deep learning and neural networks - it covers a popular and fundamental technique used in machine learning, data science and statistics: logistic regression. We cover the theory from the ground up: derivation of the solution, and applications to real-world problems. We show you how one might code their own logistic regression module in Python. This course does not require any external materials. Everything needed (Python, and some Python libraries) can be obtained for free.

Understanding the convergence performance of asynchronous stochastic gradient descent method (Async-SGD) has received increasing attention in recent years due to their foundational role in machine learning. To date, however, most of the existing works are restricted to either bounded gradient delays or convex settings. In this paper, we focus on Async-SGD and its variant Async-SGDI (which uses increasing batch size) for non-convex optimization problems with unbounded gradient delays. We prove $o(1/\sqrt{k})$ convergence rate for Async-SGD and $o(1/k)$ for Async-SGDI. Also, a unifying sufficient condition for Async-SGD's convergence is established, which includes two major gradient delay models in the literature as special cases and yields a new delay model not considered thus far.

We study the problem of off-policy evaluation (OPE) in reinforcement learning (RL), where the goal is to estimate the performance of a policy from the data generated by another policy(ies). In particular, we focus on the doubly robust (DR) estimators that consist of an importance sampling (IS) component and a performance model, and utilize the low (or zero) bias of IS and low variance of the model at the same time. Although the accuracy of the model has a huge impact on the overall performance of DR, most of the work on using the DR estimators in OPE has been focused on improving the IS part, and not much on how to learn the model. In this paper, we propose alternative DR estimators, called more robust doubly robust (MRDR), that learn the model parameter by minimizing the variance of the DR estimator. We first present a formulation for learning the DR model in RL. We then derive formulas for the variance of the DR estimator in both contextual bandits and RL, such that their gradients w.r.t. the model parameters can be estimated from the samples, and propose methods to efficiently minimize the variance. We prove that the MRDR estimators are strongly consistent and asymptotically optimal. Finally, we evaluate MRDR in bandits and RL benchmark problems, and compare its performance with the existing methods.