Type 2 diabetes mellitus (T2DM) is a chronic disease that often results in multiple complications. Risk prediction and profiling of T2DM complications is critical for healthcare professionals to design personalized treatment plans for patients in diabetes care for improved outcomes. In this paper, we study the risk of developing complications after the initial T2DM diagnosis from longitudinal patient records. We propose a novel multi-task learning approach to simultaneously model multiple complications where each task corresponds to the risk modeling of one complication. Specifically, the proposed method strategically captures the relationships (1) between the risks of multiple T2DM complications, (2) between the different risk factors, and (3) between the risk factor selection patterns. The method uses coefficient shrinkage to identify an informative subset of risk factors from high-dimensional data, and uses a hierarchical Bayesian framework to allow domain knowledge to be incorporated as priors. The proposed method is favorable for healthcare applications because in additional to improved prediction performance, relationships among the different risks and risk factors are also identified. Extensive experimental results on a large electronic medical claims database show that the proposed method outperforms state-of-the-art models by a significant margin. Furthermore, we show that the risk associations learned and the risk factors identified lead to meaningful clinical insights.

We study the local geometry of a one-hidden-layer fully-connected neural network where the training samples are generated from a multi-neuron logistic regression model. We prove that under Gaussian input, the empirical risk function employing quadratic loss exhibits strong convexity and smoothness uniformly in a local neighborhood of the ground truth, for a class of smooth activation functions satisfying certain properties, including sigmoid and tanh, as soon as the sample complexity is sufficiently large. This implies that if initialized in this neighborhood, gradient descent converges linearly to a critical point that is provably close to the ground truth without requiring a fresh set of samples at each iteration. This significantly improves upon prior results on learning shallow neural networks with multiple neurons. To the best of our knowledge, this is the first global convergence guarantee for one-hidden-layer neural networks using gradient descent over the empirical risk function without resampling at the near-optimal sampling and computational complexity.

Finding efficient decoders for quantum error correcting codes adapted to realistic experimental noise in fault-tolerant devices represents a significant challenge. In this paper we introduce several decoding algorithms complemented by deep neural decoders and apply them to analyze several fault-tolerant error correction protocols such as the surface code as well as Steane and Knill error correction. Our methods require no knowledge of the underlying noise model afflicting the quantum device making them appealing for real-world experiments. Our analysis is based on a full circuit-level noise model. It considers both distance-three and five codes, and is performed near the codes pseudo-threshold regime. Training deep neural decoders in low noise rate regimes appears to be a challenging machine learning endeavour. We provide a detailed description of our neural network architectures and training methodology. We then discuss both the advantages and limitations of deep neural decoders. Lastly, we provide a rigorous analysis of the decoding runtime of trained deep neural decoders and compare our methods with anticipated gate times in future quantum devices. Given the broad applications of our decoding schemes, we believe that the methods presented in this paper could have practical applications for near term fault-tolerant experiments.

Mild cognitive impairment (MCI) is a prodromal phase in the progression from normal aging to dementia, especially Alzheimers disease. Even though there is mild cognitive decline in MCI patients, they have normal overall cognition and thus is challenging to distinguish from normal aging. Using transcribed data obtained from recorded conversational interactions between participants and trained interviewers, and applying supervised learning models to these data, a recent clinical trial has shown a promising result in differentiating MCI from normal aging. However, the substantial amount of interactions with medical staff can still incur significant medical care expenses in practice. In this paper, we propose a novel reinforcement learning (RL) framework to train an efficient dialogue agent on existing transcripts from clinical trials. Specifically, the agent is trained to sketch disease-specific lexical probability distribution, and thus to converse in a way that maximizes the diagnosis accuracy and minimizes the number of conversation turns. We evaluate the performance of the proposed reinforcement learning framework on the MCI diagnosis from a real clinical trial. The results show that while using only a few turns of conversation, our framework can significantly outperform state-of-the-art supervised learning approaches.

Training complex machine learning models for prediction often requires a large amount of data that is not always readily available. Leveraging these external datasets from related but different sources is therefore an important task if good predictive models are to be built for deployment in settings where data can be rare. In this paper we propose a novel approach to the problem in which we use multiple GAN architectures to learn to translate from one dataset to another, thereby allowing us to effectively enlarge the target dataset, and therefore learn better predictive models than if we simply used the target dataset. We show the utility of such an approach, demonstrating that our method improves the prediction performance on the target domain over using just the target dataset and also show that our framework outperforms several other benchmarks on a collection of real-world medical datasets.

We propose a hybrid algorithm for top-$n$ recommendation task that allows to incorporate both user and item side information within the standard collaborative filtering approach. The algorithm extends PureSVD -- one of the state-of-the-art latent factor models -- by exploiting a generalized formulation of the singular value decomposition. This allows to inherit key advantages of the classical algorithm such as highly efficient Lanczos-based optimization procedure, minimal parameter tuning during a model selection phase and a quick folding-in computation to generate recommendations instantly even in a highly dynamic online environment. Within the generalized formulation itself we provide an efficient scheme for side information fusion which avoids undesirable computational overhead and addresses the scalability question. Evaluation of the model is performed in both standard and cold-start scenarios using the datasets with different sparsity levels. We demonstrate in which cases our approach outperforms conventional methods and also provide some intuition on when it may give no significant improvement.

In this paper we consider online mirror descent (OMD) algorithms, a class of scalable online learning algorithms exploiting data geometric structures through mirror maps. Necessary and sufficient conditions are presented in terms of the step size sequence $\{\eta_t\}_{t}$ for the convergence of an OMD algorithm with respect to the expected Bregman distance induced by the mirror map. The condition is $\lim_{t\to\infty}\eta_t=0, \sum_{t=1}^{\infty}\eta_t=\infty$ in the case of positive variances. It is reduced to $\sum_{t=1}^{\infty}\eta_t=\infty$ in the case of zero variances for which the linear convergence may be achieved by taking a constant step size sequence. A sufficient condition on the almost sure convergence is also given. We establish tight error bounds under mild conditions on the mirror map, the loss function, and the regularizer. Our results are achieved by some novel analysis on the one-step progress of the OMD algorithm using smoothness and strong convexity of the mirror map and the loss function.

The trace of matrix functions, often called spectral sums, e.g., rank, log-determinant and nuclear norm, appear in many machine learning tasks. However, optimizing or computing such (parameterized) spectral sums typically involves the matrix decomposition at the cost cubic in the matrix dimension, which is expensive for large-scale applications. Several recent works were proposed to approximate large-scale spectral sums utilizing polynomial function approximations and stochastic trace estimators. However, all prior works on this line have studied biased estimators, and their direct adaptions to an optimization task under stochastic gradient descent (SGD) frameworks often do not work as accumulated biased errors prevent stable convergence to the optimum. To address the issue, we propose the provable optimal unbiased estimator by randomizing Chebyshev polynomial degrees. We further introduce two additional techniques for accelerating SGD, where key ideas are on sharing randomness among many estimations during the iterative procedure. Finally, we showcase two applications of the proposed SGD schemes: matrix completion and learning Gaussian process, under the real-world datasets.

We propose a novel distributed inference algorithm for continuous graphical models, by extending Stein variational gradient descent (SVGD) (Liu & Wang, 2016) to leverage the Markov dependency structure of the distribution of interest. Our approach combines SVGD with a set of structured local kernel functions defined on the Markov blanket of each node, which alleviates the curse of high dimensionality and simultaneously yields a distributed algorithm for decentralized inference tasks. We justify our method with theoretical analysis and show that the use of local kernels can be viewed as a new type of localized approximation that matches the target distribution on the conditional distributions of each node over its Markov blanket. Our empirical results show that our method outperforms a variety of baselines including standard MCMC and particle message passing methods.

We propose an efficient stochastic variational approach to GP classification building on Polya- Gamma data augmentation and inducing points, which is based on closed-form updates of natural gradients. We evaluate the algorithm on real-world datasets containing up to 11 million data points and demonstrate that it is up to three orders of magnitude faster than the state-of-the-art while being competitive in terms of prediction performance.