Deep learning is a fascinating sub field of machine learning that creates artificially intelligent systems inspired by the structure and function of the brain. The basis of these models are bio-inspired artificial neural networks that mimic the neural connectivity of animal brains to carry out cognitive functions such as problem solving. A field with the most impressive results of neuromorphic computing is that of visual image analysis. Similar to how our brains learn to recognize objects in order to make predictions and act upon them, artificial intelligence must be shown millions of pictures before they are able to generalize them in order to make their best educated guesses for images they have never seen before. Professor Cheol Seong Hwang from the Department of Material Science and Engineering at Seoul National University and his research team have developed a method to accelerate the image recognition process by combining the inherent efficiency of resistive random access memory (ReRAM) and cross-bar array structures, two of the most commonly used hardware. Many of us have performed a reversed image search to find information based on a certain image in order to browse similar results.

Preparing employees for jobs of the future will require leaders in business, government, and higher education to work together. That was a major takeaway from a conference at Northeastern's Toronto campus to discuss the results of a Northeastern-Gallup poll on attitudes toward artificial intelligence in the U.S., the U.K and Canada. "We have to do more partnering with universities on that note," Helena Gottschling, the chief human resources officer at the Royal Bank of Canada, told an audience comprised of the three sectors gathered for a conference at Northeastern's Toronto campus this week. She added that employers need to do more to communicate "what we need in our workforce through the universities, so that we're helping to inform the skills and capabilities that the universities are growing through the student populations." The conference follows the publication of a survey conducted by Northeastern and Gallup that revealed an international cross-section of opinions about artificial intelligence as economies around the world undergo the transformative move to automation.

The third annual Modern Artificial Intelligence (AI) seminar series at NYU Tandon, bringing together students and experts to discuss recent advances in the field, wrapped up on December 6 with a presentation by Raia Hadsell, Head of Robotics Research at DeepMind. In the final presentation of the series, sponsored by the Department of Electrical and Computer Engineering and organized by Professor Anna Choromanska, Hadsell explored ways in which human learning can inform machine learning systems to develop highly sophisticated AI to solve complex real-world tasks. The Fall roster kicked off in early October with a lecture by Facebook AI Research's Leon Bottou. The researcher, who harbors the long-term ambition of replicating human-level intelligence, examined causal inference, or finding the relationship between existing facts and objects. Next, on November 14, Francis Bach, researcher at Institut National de Recherche en Informatique et en Automatique (INRIA) in France, spoke about a new generation of "distributed optimization" schemes that are critically needed to scale algorithms to massive data.

Artificial Intelligence is generally seen as a great advance and benefit for mankind. Smart humans, however, see it as our undoing and even possibly our extermination. Much of modern technology has far more prospect for harm than for good. There are a large number of technologies that impose massive costs on life on Earth. The article below from Crime & Power, "A Hard Look At Artificial intelligence" by Jerry Day, sounds like a horror science fiction story.

A new research project in Australia is using motion detectors and muscle sensors to track sheep shearers in an effort to minimize on the-job-injuries. Sheep shearers are six times more likely to be injured in the workplace than the average Australian worker. Data from sensors attached to sheep shearers will be used to model worker movement throughout the workday and test new ways of doing the job without risking injury. The study, a joint project between University of Melbourne and the trade group Australian Wool Innovation, uses sensors to measure electrical activity in muscles. These sensors are placed directly on the skin of the lower back and upper thighs, the ABC reported, while motion detectors are placed around the joints to track a worker's posture and shearing motions.

Bayesian hybrid models fuse physics-based insights with machine learning constructs to correct for systematic bias. In this paper, we compare Bayesian hybrid models against physics-based glass-box and Gaussian process black-box surrogate models. We consider ballistic firing as an illustrative case study for a Bayesian decision-making workflow. First, Bayesian calibration is performed to estimate model parameters. We then use the posterior distribution from Bayesian analysis to compute optimal firing conditions to hit a target via a single-stage stochastic program. The case study demonstrates the ability of Bayesian hybrid models to overcome systematic bias from missing physics with less data than the pure machine learning approach. Ultimately, we argue Bayesian hybrid models are an emerging paradigm for data-informed decision-making under parametric and epistemic uncertainty.

While meta-learning approaches that utilize neural network representations have made progress in few-shot image classification, reinforcement learning, and, more recently, image semantic segmentation, the training algorithms and model architectures have become increasingly specialized to the few-shot domain. A natural question that arises is how to develop learning systems that scale from few-shot to many-shot settings while yielding competitive performance in both. One scalable potential approach that does not require ensembling many models nor the computational costs of relation networks, is to meta-learn an initialization. In this work, we study first-order meta-learning of initializations for deep neural networks that must produce dense, structured predictions given an arbitrary amount of training data for a new task. Our primary contributions include (1), an extension and experimental analysis of first-order model agnostic meta-learning algorithms (including FOMAML and Reptile) to image segmentation, (2) a novel neural network architecture built for parameter efficiency and fast learning which we call EfficientLab, (3) a formalization of the generalization error of meta-learning algorithms, which we leverage to decrease error on unseen tasks, and (4) a small benchmark dataset, FP-k, for the empirical study of how meta-learning systems perform in both few- and many-shot settings. We show that meta-learned initializations for image segmentation provide value for both canonical few-shot learning problems and larger datasets, outperforming ImageNet-trained initializations for up to 400 densely labeled examples. We find that our network, with an empirically estimated optimal update procedure, yields state of the art results on the FSS-1000 dataset while only requiring one forward pass through a single model at evaluation time.

Neural Networks and Random Forests: two popular techniques for supervised learning that are seemingly disconnected in their formulation and optimization method, have recently been linked in a single construct. The connection pivots on assembling an artificial Neural Network with nodes that allow for a gate-like function to mimic a tree split, optimized using the standard approach of recursively applying the chain rule to update its parameters. Yet two main challenges have impeded wide use of this hybrid approach: \emph{(a)} the inability of global gradient descent techniques to optimize hierarchical parameters (as introduced by the gate function); and \emph{(b)} the construction of the tree structure, which has relied on standard decision tree algorithms to learn the network topology or incrementally (and heuristically) searching the space at random. We propose a probabilistic construct that exploits the idea of a node's \emph{unexplained potential} (the total error channeled through the node) in order to decide where to expand further, mimicking the standard tree construction in a Neural Network setting, alongside a modified gradient descent that first locally optimizes an expanded node before a global optimization. The probabilistic approach allows us to evaluate each new split as a ratio of likelihoods that balance the statistical improvement in explaining the evidence against the additional model complexity --- thus providing a natural stopping condition. The result is a novel classification and regression technique that leverages the strength of both: a tree-structure that grows naturally and is simple to interpret with the plasticity of Neural Networks that allow for soft margins and slanted boundaries.

Time series regression analysis in econometrics typically involves a framework relying on a set of mixing conditions to establish consistency and asymptotic normality of parameter estimates and HAC-type estimators of the residual long-run variances to conduct proper inference. This article introduces structured machine learning regressions for high-dimensional time series data using the aforementioned commonly used setting. To recognize the time series data structures we rely on the sparse-group LASSO estimator. We derive a new Fuk-Nagaev inequality for a class of $\tau$-dependent processes with heavier than Gaussian tails, nesting $\alpha$-mixing processes as a special case, and establish estimation, prediction, and inferential properties, including convergence rates of the HAC estimator for the long-run variance based on LASSO residuals. An empirical application to nowcasting US GDP growth indicates that the estimator performs favorably compared to other alternatives and that the text data can be a useful addition to more traditional numerical data.

We propose a communication- and computation-efficient distributed optimization algorithm using second-order information for solving empirical risk minimization (ERM) problems with a nonsmooth regularization term. Our algorithm is applicable to both the primal and the dual ERM problem. Current second-order and quasi-Newton methods for this problem either do not work well in the distributed setting or work only for specific regularizers. Our algorithm uses successive quadratic approximations of the smooth part, and we describe how to maintain an approximation of the (generalized) Hessian and solve subproblems efficiently in a distributed manner. When applied to the distributed dual ERM problem, unlike state of the art that takes only the block-diagonal part of the Hessian, our approach is able to utilize global curvature information and is thus magnitudes faster. The proposed method enjoys global linear convergence for a broad range of non-strongly convex problems that includes the most commonly used ERMs, thus requiring lower communication complexity. It also converges on non-convex problems, so has the potential to be used on applications such as deep learning. Computational results demonstrate that our method significantly improves on communication cost and running time over the current state-of-the-art methods.