AI market revenues are expected to grow 487% within five years, as nearly every industry looks for ways to embed AI technology into enterprise applications and processes, according to a recent IDC study. But companies need AI talent to make it all happen. Demand for AI experts is growing, but a shortage of skilled professionals can prevent companies from taking full advantage of the advanced technology. A study released by Tech Pro Research last year found more than 40% of respondents said their enterprise IT personnel don't have the skills required to implement and support AI and machine learning. Ng isn't the only one looking to train AI experts.

About this course: This course covers advanced topics in R programming that are necessary for developing powerful, robust, and reusable data science tools. Topics covered include functional programming in R, robust error handling, object oriented programming, profiling and benchmarking, debugging, and proper design of functions. Upon completing this course you will be able to identify and abstract common data analysis tasks and to encapsulate them in user-facing functions. Because every data science environment encounters unique data challenges, there is always a need to develop custom software specific to your organization's mission. You will also be able to define new data types in R and to develop a universe of functionality specific to those data types to enable cleaner execution of data science tasks and stronger reusability within a team.

In the 1950s, scientists began working to build machines capable of imitating intelligent human behaviour. Progress accelerated recently when we entered a new phase of machine learning--one that has led to a dramatic decrease in the cost of prediction. Moore's Law (the 18-month doubling of transistor intensity on microprocessors) was the driver of the previous phase--a revolution in technology hardware that advanced mobile innovation and smartphone adoption. Increased connectivity and scalable cloud-based storage catalysed a step change in the amount of data we collected and consumed. Information taken from sensors, images, videos, and other digital sources is being used to generate a more accurate view of real-time context.

About this course: Welcome to the Advanced Linear Models for Data Science Class 1: Least Squares. This class is an introduction to least squares from a linear algebraic and mathematical perspective. Before beginning the class make sure that you have the following: - A basic understanding of linear algebra and multivariate calculus. After taking this course, students will have a firm foundation in a linear algebraic treatment of regression modeling. This will greatly augment applied data scientists' general understanding of regression models.

About this course: Probabilistic graphical models (PGMs) are a rich framework for encoding probability distributions over complex domains: joint (multivariate) distributions over large numbers of random variables that interact with each other. These representations sit at the intersection of statistics and computer science, relying on concepts from probability theory, graph algorithms, machine learning, and more. They are the basis for the state-of-the-art methods in a wide variety of applications, such as medical diagnosis, image understanding, speech recognition, natural language processing, and many, many more. They are also a foundational tool in formulating many machine learning problems. This course is the third in a sequence of three.

Gaussian process (GP) regression is a powerful interpolation technique due to its flexibility in capturing non-linearity. In this paper, we provide a general framework for understanding the frequentist coverage of point-wise and simultaneous Bayesian credible sets in GP regression. As an intermediate result, we develop a Bernstein von-Mises type result under supremum norm in random design GP regression. Identifying both the mean and covariance function of the posterior distribution of the Gaussian process as regularized $M$-estimators, we show that the sampling distribution of the posterior mean function and the centered posterior distribution can be respectively approximated by two population level GPs. By developing a comparison inequality between two GPs, we provide exact characterization of frequentist coverage probabilities of Bayesian point-wise credible intervals and simultaneous credible bands of the regression function. Our results show that inference based on GP regression tends to be conservative; when the prior is under-smoothed, the resulting credible intervals and bands have minimax-optimal sizes, with their frequentist coverage converging to a non-degenerate value between their nominal level and one. As a byproduct of our theory, we show that the GP regression also yields minimax-optimal posterior contraction rate relative to the supremum norm, which provides a positive evidence to the long standing problem on optimal supremum norm contraction rate in GP regression.

Accurate diagnosis of Alzheimer's Disease (AD) entails clinical evaluation of multiple cognition metrics and biomarkers. Metrics such as the Alzheimer's Disease Assessment Scale - Cognitive test (ADAS-cog) comprise multiple subscores that quantify different aspects of a patient's cognitive state such as learning, memory, and language production/comprehension. Although computer-aided diagnostic techniques for classification of a patient's current disease state exist, they provide little insight into the relationship between changes in brain structure and different aspects of a patient's cognitive state that occur over time in AD. We have developed a Convolutional Neural Network architecture that can concurrently predict the trajectories of the 13 subscores comprised by a subject's ADAS-cog examination results from a current minimally preprocessed structural MRI scan up to 36 months from image acquisition time without resorting to manual feature extraction. Mean performance metrics are within range of those of existing techniques that require manual feature selection and are limited to predicting aggregate scores.

Bipartite matching, where agents on one side of a market are matched to agents or items on the other, is a classical problem in computer science and economics, with widespread application in healthcare, education, advertising, and general resource allocation. A practitioner's goal is typically to maximize a matching market's economic efficiency, possibly subject to some fairness requirements that promote equal access to resources. A natural balancing act exists between fairness and efficiency in matching markets, and has been the subject of much research. In this paper, we study a complementary goal-- balancing diversity and efficiency--in a generalization of bipartite matching where agents on one side of the market can be matched to sets of agents on the other. Adapting a classical definition of the diversity of a set, we propose a quadratic programming-based approach to solving a supermodular minimization problem that balances diversity and total weight of the solution. We also provide a scalable greedy algorithm with theoretical performance bounds. We then define the price of diversity, a measure of the efficiency loss due to enforcing diversity, and give a worst-case theoretical bound. Finally, we demonstrate the efficacy of our methods on three real-world datasets, and show that the price of diversity is not bad in practice. Our code is publicly accessible for further research.

Missing data and noisy observations pose significant challenges for reliably predicting events from irregularly sampled multivariate time series (longitudinal) data. Imputation methods, which are typically used for completing the data prior to event prediction, lack a principled mechanism to account for the uncertainty due to missingness. Alternatively, state-of-the-art joint modeling techniques can be used for jointly modeling the longitudinal and event data and compute event probabilities conditioned on the longitudinal observations. These approaches, however, make strong parametric assumptions and do not easily scale to multivariate signals with many observations. Our proposed approach consists of several key innovations. First, we develop a flexible and scalable joint model based upon sparse multiple-output Gaussian processes. Unlike state-of-the-art joint models, the proposed model can explain highly challenging structure including non-Gaussian noise while scaling to large data. Second, we derive an optimal policy for predicting events using the distribution of the event occurrence estimated by the joint model. The derived policy trades-off the cost of a delayed detection versus incorrect assessments and abstains from making decisions when the estimated event probability does not satisfy the derived confidence criteria. Experiments on a large dataset show that the proposed framework significantly outperforms state-of-the-art techniques in event prediction.

It is well established that neural networks with deep architectures perform better than shallow networks for many tasks in machine learning. In statistical physics, while there has been recent interest in representing physical data with generative modelling, the focus has been on shallow neural networks. A natural question to ask is whether deep neural networks hold any advantage over shallow networks in representing such data. We investigate this question by using unsupervised, generative graphical models to learn the probability distribution of a two-dimensional Ising system. Deep Boltzmann machines, deep belief networks, and deep restricted Boltzmann networks are trained on thermal spin configurations from this system, and compared to the shallow architecture of the restricted Boltzmann machine. We benchmark the models, focussing on the accuracy of generating energetic observables near the phase transition, where these quantities are most difficult to approximate. Interestingly, after training the generative networks, we observe that the accuracy essentially depends only on the number of neurons in the first hidden layer of the network, and not on other model details such as network depth or model type. This is evidence that shallow networks are more efficient than deep networks at representing physical probability distributions associated with Ising systems near criticality.