The former is useful for tasks such as planning and control, and the latter for classification. The automatic extraction of this information is challenging because of the high-dimensionality and entangled encoding inherent to the image representation. This article introduces two theoretical approaches aimed at the resolution of these challenges. The approaches allow for the interpolation and extrapolation of images from an image sequence by joint estimation of the image representation and the generators of the sequence dynamics. In the first approach, the image representations are learned using probabilistic PCA . The linear-Gaussian conditional distributions allow for a closed form analytical description of the latent distributions but assumes the underlying image manifold is a linear subspace. In the second approach, the image representations are learned using probabilistic nonlinear PCA which relieves the linear manifold assumption at the cost of requiring a variational approximation of the latent distributions. In both approaches, the underlying dynamics of the image sequence are modelled explicitly to disentangle them from the image representations. The dynamics themselves are modelled with Lie group structure which enforces the desirable properties of smoothness and composability of inter-image transformations.
Research in the field of traditional computing systems is slowing down, with new types of computing moving to the forefront now. A team of engineers from Pennsylvania State University (Penn State) in the U.S. has been working on creating a type of computing based on our brain's neural networks' systems all while using the brain's analog nature. The team has discovered that graphene-based memory resistors show promise for this new computing form. Their findings were recently published in Nature Communications. "We have powerful computers, no doubt about that, the problem is you have to store the memory in one place and do the computing somewhere else," said Saptarshi Das, the team leader and Penn State assistant professor of engineering science and mechanics.
The team leveraged advanced analytics tools to create cost-effective bus routes that allow nonprofit organizations to deliver meals to senior citizens, as well as K-12 students and families who would otherwise rely on schools for free meals. The machine learning tools identified ideal distribution locations to reach as many people as possible, three days a week. The program began in July, and nearly 6,000 meals are delivered each month. It has since expanded to include dinners that feed a family of four. Carnegie Mellon's Metro21: Smart City Institute leads the project alongside Allies for Children, United Way of Southwestern Pennsylvania and the Greater Pittsburgh Community Food Bank.
Kusne, A. Gilad, Yu, Heshan, Wu, Changming, Zhang, Huairuo, Hattrick-Simpers, Jason, DeCost, Brian, Sarker, Suchismita, Oses, Corey, Toher, Cormac, Curtarolo, Stefano, Davydov, Albert V., Agarwal, Ritesh, Bendersky, Leonid A., Li, Mo, Mehta, Apurva, Takeuchi, Ichiro
Active learning - the field of machine learning (ML) dedicated to optimal experiment design, has played a part in science as far back as the 18th century when Laplace used it to guide his discovery of celestial mechanics . In this work we focus a closed-loop, active learning-driven autonomous system on another major challenge, the discovery of advanced materials against the exceedingly complex synthesis-processes-structure-property landscape. We demonstrate autonomous research methodology (i.e. autonomous hypothesis definition and evaluation) that can place complex, advanced materials in reach, allowing scientists to fail smarter, learn faster, and spend less resources in their studies, while simultaneously improving trust in scientific results and machine learning tools. Additionally, this robot science enables science-over-the-network, reducing the economic impact of scientists being physically separated from their labs. We used the real-time closed-loop, autonomous system for materials exploration and optimization (CAMEO) at the synchrotron beamline to accelerate the fundamentally interconnected tasks of rapid phase mapping and property optimization, with each cycle taking seconds to minutes, resulting in the discovery of a novel epitaxial nanocomposite phase-change memory material.
This paper studies the high-dimensional mixed linear regression (MLR) where the output variable comes from one of the two linear regression models with an unknown mixing proportion and an unknown covariance structure of the random covariates. Building upon a high-dimensional EM algorithm, we propose an iterative procedure for estimating the two regression vectors and establish their rates of convergence. Based on the iterative estimators, we further construct debiased estimators and establish their asymptotic normality. For individual coordinates, confidence intervals centered at the debiased estimators are constructed. Furthermore, a large-scale multiple testing procedure is proposed for testing the regression coefficients and is shown to control the false discovery rate (FDR) asymptotically. Simulation studies are carried out to examine the numerical performance of the proposed methods and their superiority over existing methods. The proposed methods are further illustrated through an analysis of a dataset of multiplex image cytometry, which investigates the interaction networks among the cellular phenotypes that include the expression levels of 20 epitopes or combinations of markers.
While deep learning has resulted in major breakthroughs in many application domains, the frameworks commonly used in deep learning remain fragile to artificially-crafted and imperceptible changes in the data. In response to this fragility, adversarial training has emerged as a principled approach for enhancing the robustness of deep learning with respect to norm-bounded perturbations. However, there are other sources of fragility for deep learning that are arguably more common and less thoroughly studied. Indeed, natural variation such as lighting or weather conditions can significantly degrade the accuracy of trained neural networks, proving that such natural variation presents a significant challenge for deep learning. In this paper, we propose a paradigm shift from perturbation-based adversarial robustness toward model-based robust deep learning. Our objective is to provide general training algorithms that can be used to train deep neural networks to be robust against natural variation in data. Critical to our paradigm is first obtaining a model of natural variation which can be used to vary data over a range of natural conditions. Such models may be either known a priori or else learned from data. In the latter case, we show that deep generative models can be used to learn models of natural variation that are consistent with realistic conditions. We then exploit such models in three novel model-based robust training algorithms in order to enhance the robustness of deep learning with respect to the given model. Our extensive experiments show that across a variety of naturally-occurring conditions and across various datasets, deep neural networks trained with our model-based algorithms significantly outperform both standard deep learning algorithms as well as norm-bounded robust deep learning algorithms.
A massive problem like Alzheimer's disease (AD) — which affects nearly 50 million people worldwide — requires bold solutions. New funding expected to total $17.8 million, awarded to the Keck School of Medicine of USC's Mark and Mary Stevens Neuroimaging and Informatics Institute (INI) and its collaborators, is one key piece of that puzzle. The five-year National Institutes of Health (NIH)-funded effort, "Ultrascale Machine Learning to Empower Discovery in Alzheimer's Disease Biobanks," known as AI4AD, will develop state-of-the-art artificial intelligence (AI) methods and apply them to giant databases of genetic, imaging and cognitive data collected from AD patients. Forty co-investigators at 11 research centers will team up to leverage AI and machine learning to bolster precision diagnostics, prognosis and the development of new treatments for AD. "Our team of experts in computer science, genetics, neuroscience and imaging sciences will create algorithms that analyze data at a previously impossible scale," said Paul Thompson, PhD, associate director of the INI and project leader for the new grant. "Collectively, this will enable the discovery of new features in the genome that influence the biological processes involved in Alzheimer's disease." Predicting a diagnosis The project's first objective is to identify genetic and biological markers that predict an AD diagnosis — and to distinguish between several subtypes of the disease. To accomplish this, the research team will apply sophisticated AI and machine learning methods to a variety of data types, including tens of thousands of brain images and whole genome sequences. The investigators then will relate these findings to the clinical progression of AD, including in patients who have not yet developed dementia symptoms. The researchers will train AI methods on large databases of brain scans to identify patterns that can help detect the disease as it emerges in individual patients. "As we get older, each of us has a unique mix of brain changes that occur for decades before we develop any signs of Alzheimer's disease — changes in our blood vessels, the buildup of abnormal protein deposits and brain cell loss," said Thompson, who also directs INI's Imaging Genetics Center. "Our new AI methods will help us determine what changes are happening in each patient, as well as drivers of these processes in their DNA, that we can target with new drugs." The team is even creating a dedicated "Drug Repurposing Core" to identify ways to repurpose existing drugs to target newly identified segments of the genome, molecules or neurobiological processes involved in the disease. "We predict that combining AI with whole genome data and advanced brain scans will outperform methods used today to predict Alzheimer's disease progression," Thompson said. Advancing AI The AI4AD effort is part of the "Cognitive Systems Analysis of Alzheimer's Disease Genetic and Phenotypic Data" and "Harmonization of Alzheimer's Disease and Related Dementias (AD/ADRD) Genetic, Epidemiologic, and Clinical Data to Enhance Therapeutic Target Discovery" initiatives from the NIH's National Institute on Aging. These initiatives aim to create and develop advanced AI methods and apply them to extensive and harmonized rich genomic, imaging and cognitive data. Collectively, the goals of AI4AD leverage the promise of machine learning to contribute to precision diagnostics, prognostication, and targeted and novel treatments. Thompson and his USC team will collaborate with four co-principal investigators at the University of Pennsylvania, the University of Pittsburgh and the Indiana University School of Medicine. The researchers will also host regular training events at major AD neuroimaging and genetics conferences to help disseminate newly developed AI tools to investigators across the field. Research reported in this publication will be supported by the National Institute on Aging of the National Institutes of Health under Award Number U01AG068057. Also involved in the project are INI faculty members Neda Jahanshad and Lauren Salminen, as well as consortium manager Sophia Thomopoulos. — Zara Greenbaum
We study two-sided decentralized matching markets in which participants have uncertain preferences. We present a statistical model to learn the preferences. The model incorporates uncertain state and the participants' competition on one side of the market. We derive an optimal strategy that maximizes the agent's expected payoff and calibrate the uncertain state by taking the opportunity costs into account. We discuss the sense in which the matching derived from the proposed strategy has a stability property. We also prove a fairness property that asserts that there exists no justified envy according to the proposed strategy. We provide numerical results to demonstrate the improved payoff, stability and fairness, compared to alternative methods.
Effective training of advanced ML models requires large amounts of labeled data, which is often scarce in scientific problems given the substantial human labor and material cost to collect labeled data. This poses a challenge on determining when and where we should deploy measuring instruments (e.g., in-situ sensors) to collect labeled data efficiently. This problem differs from traditional pool-based active learning settings in that the labeling decisions have to be made immediately after we observe the input data that come in a time series. In this paper, we develop a real-time active learning method that uses the spatial and temporal contextual information to select representative query samples in a reinforcement learning framework. To reduce the need for large training data, we further propose to transfer the policy learned from simulation data which is generated by existing physics-based models. We demonstrate the effectiveness of the proposed method by predicting streamflow and water temperature in the Delaware River Basin given a limited budget for collecting labeled data. We further study the spatial and temporal distribution of selected samples to verify the ability of this method in selecting informative samples over space and time.
It is well known that machine learning methods can be vulnerable to adversarially-chosen perturbations of their inputs. Despite significant progress in the area, foundational open problems remain. Here we address several of these key questions. We derive exact and approximate Bayes-optimal robust classifiers for the important setting of two- and three-class Gaussian classification problems with arbitrary imbalance, for $\ell_2$ and $\ell_\infty$ adversaries. In contrast to classical Bayes-optimal classifiers, decisions here cannot be made pointwise and new theoretical approaches are needed. We develop and leverage new tools, including recent breakthroughs from probability theory on robust isoperimetry (Cianci et al, 2011, Mossel and Neeman 2015), which, to our knowledge, have not yet been used in the area. Our results reveal tradeoffs between standard and robust accuracy that grow when data is imbalanced. We also show further foundational results, including an analysis of the loss landscape, classification calibration for convex losses in certain models, and finite sample rates for the robust risk.