Rapid progress in machine learning and artificial intelligence (AI) has brought increasing attention to the potential impacts of AI technologies on society. In this paper we discuss one such potential impact: the problem of accidents in machine learning systems, defined as unintended and harmful behavior that may emerge from poor design of real-world AI systems. We present a list of five practical research problems related to accident risk, categorized according to whether the problem originates from having the wrong objective function ("avoiding side effects" and "avoiding reward hacking"), an objective function that is too expensive to evaluate frequently ("scalable supervision"), or undesirable behavior during the learning process ("safe exploration" and "distributional shift"). We review previous work in these areas as well as suggesting research directions with a focus on relevance to cutting-edge AI systems. Finally, we consider the high-level question of how to think most productively about the safety of forward-looking applications of AI.

Concrete Problems in AI Safety

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of-the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research labs. In a user study, subject matter experts build models 2.8X faster and increase predictive performance an average 45.5% versus seven hours of hand labeling. We study the modeling tradeoffs in this new setting and propose an optimizer for automating tradeoff decisions that gives up to 1.8X speedup per pipeline execution. In two collaborations, with the U.S. Department of Veterans Affairs and the U.S. Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides 132% average improvements to predictive performance over prior heuristic approaches and comes within an average 3.60% of the predictive performance of large hand-curated training sets.

/pdf/snorkel-rapid-training-data-creation-with-weak-supervision-15qyryp476.pdf

Snorkel: rapid training data creation with weak supervision

Recent years have witnessed an explosion in the volume and variety of data collected in all scientific disciplines and industrial settings. Such massive data sets present a number of challenges to researchers in statistics and machine learning. This book provides a self-contained introduction to the area of high-dimensional statistics, aimed at the first-year graduate level. It includes chapters that are focused on core methodology and theory - including tail bounds, concentration inequalities, uniform laws and empirical process, and random matrices - as well as chapters devoted to in-depth exploration of particular model classes - including sparse linear models, matrix models with rank constraints, graphical models, and various types of non-parametric models. With hundreds of worked examples and exercises, this text is intended both for courses and for self-study by graduate students and researchers in statistics, machine learning, and related fields who must understand, apply, and adapt modern statistical methods suited to large-scale data.

High-Dimensional Statistics: A Non-Asymptotic Viewpoint

electrical and computer engineering ece courses ece 257a multiuser communication systems 4 congestion control convex programming and dual controller fair end end rate allocation max min fair vs proportional, electrical systems engineering washington university arye nehorai eugene and martha lohman professor of electrical engineering phd stanford university signal processing imaging biomedicine communications, ieee transactions on aerospace and electronic systems ieee transactions on aerospace and electronic systems focuses on the organization design development integration and operation of complex systems for space air, department of electrical engineering and computer science h kumar wickramsinghe department chair 2213 engineering hall 949 824 4821 http www eng uci edu dept eecs overview electrical engineering and computer science is, download electrical and electronics engineering ebooks syst mes temps discret commande num rique des proc d s pdf 499 ko terminology and symbols in control engineering pdf 326 ko the best of thomas, publications stream wise list iit kanpur papers published in journals in 2016 dutta s patchaikani p k behera l near optimal controller for nonlinear continuous time systems with unknown dynamics, resolve a doi name type or paste a doi name into the text box click go your browser will take you to a web page url associated with that doi name send questions or comments to doi, peer reviewed journal ijera com international journal of engineering research and applications ijera is an open access online peer reviewed international journal that publishes research, dod sbir 2016 2 sbir gov note the solicitations and topics listed on this site are copies from the various sbir agency solicitations and are not necessarily the latest and most up, an english japanese dictionary of electrical engineering c 2952 9 691 c band c c contact c c maccs centre for mathematical modelling and computer simulation, the of and to a in that is was he for it with as his on be most common text click on the icon to return to www berro com and to enjoy and benefit the of and to a in that is was he for it with as his on be at by i this had

Detection, Estimation, and Modulation Theory

We show that gradient descent converges to a local minimizer, almost surely with random initialization. This is proved by applying the Stable Manifold Theorem from dynamical systems theory.

/pdf/gradient-descent-only-converges-to-minimizers-3kgwji23wa.pdf

Gradient Descent Only Converges to Minimizers

Crowdsourcing is a popular paradigm for effectively collecting labels at low cost. The Dawid-Skene estimator has been widely used for inferring the true labels from the noisy labels provided by non-expert crowdsourcing workers. However, since the estimator maximizes a non-convex log-likelihood function, it is hard to theoretically justify its performance. In this paper, we propose a two-stage efficient algorithm for multi-class crowd labeling problems. The first stage uses the spectral method to obtain an initial estimate of parameters. Then the second stage refines the estimation by optimizing the objective function of the Dawid-Skene estimator via the EM algorithm. We show that our algorithm achieves the optimal convergence rate up to a logarithmic factor. We conduct extensive experiments on synthetic and real datasets. Experimental results demonstrate that the proposed algorithm is comparable to the most accurate empirical approach, while outperforming several other recently proposed methods.

Spectral Methods meet EM: A Provably Optimal Algorithm for Crowdsourcing

This paper addresses the problem of unsupervised domain adaption from theoretical and algorithmic perspectives. Existing domain adaptation theories naturally imply minimax optimization algorithms, which connect well with the domain adaptation methods based on adversarial learning. However, several disconnections still exist and form the gap between theory and algorithm. We extend previous theories (Mansour et al., 2009c; Ben-David et al., 2010) to multiclass classification in domain adaptation, where classifiers based on the scoring functions and margin loss are standard choices in algorithm design. We introduce Margin Disparity Discrepancy, a novel measurement with rigorous generalization bounds, tailored to the distribution comparison with the asymmetric margin loss, and to the minimax optimization for easier training. Our theory can be seamlessly transformed into an adversarial learning algorithm for domain adaptation, successfully bridging the gap between theory and algorithm. A series of empirical studies show that our algorithm achieves the state of the art accuracies on challenging domain adaptation tasks.

Bridging Theory and Algorithm for Domain Adaptation

/pdf/bridging-theory-and-algorithm-for-domain-adaptation-3hmr3o9lqp.pdf

Bridging Theory and Algorithm for Domain Adaptation.

The Dawid-Skene estimator has been widely used for inferring the true labels from the noisy labels provided by non-expert crowdsourcing workers. However, since the estimator maximizes a non-convex log-likelihood function, it is hard to theoretically justify its performance. In this paper, we propose a two-stage efficient algorithm for multi-class crowd labeling problems. The first stage uses the spectral method to obtain an initial estimate of parameters. Then the second stage refines the estimation by optimizing the objective function of the Dawid-Skene estimator via the EM algorithm. We show that our algorithm achieves the optimal convergence rate up to a logarithmic factor. We conduct extensive experiments on synthetic and real datasets. Experimental results demonstrate that the proposed algorithm is comparable to the most accurate empirical approach, while outperforming several other recently proposed methods.

/pdf/spectral-methods-meet-em-a-provably-optimal-algorithm-for-39copuz3fg.pdf

/pdf/spectral-methods-meet-em-a-provably-optimal-algorithm-for-1awe4akiep.pdf

Yuchen Zhang

Papers

Spectral Methods meet EM: A Provably Optimal Algorithm for Crowdsourcing

Bridging Theory and Algorithm for Domain Adaptation

Bridging Theory and Algorithm for Domain Adaptation.

Spectral Methods meet EM: A Provably Optimal Algorithm for Crowdsourcing

Spectral methods meet EM: a provably optimal algorithm for crowdsourcing