In high stakes applications such as healthcare and finance analytics, the interpretability of predictive models is required and necessary for domain practitioners to trust the predictions. Traditional machine learning models, e.g., logistic regression (LR), are easy to interpret in nature. However, many of these models aggregate time-series data without considering the temporal correlations and variations. Therefore, their performance cannot match up to recurrent neural network (RNN) based models, which are nonetheless difficult to interpret. In this paper, we propose a general framework TRACER to facilitate accurate and interpretable predictions, with a novel model TITV devised for healthcare analytics and other high stakes applications such as financial investment and risk management. Different from LR and other existing RNN-based models, TITV is designed to capture both the time-invariant and the time-variant feature importance using a feature-wise transformation subnetwork and a self-attention subnetwork, for the feature influence shared over the entire time series and the time-related importance respectively. Healthcare analytics is adopted as a driving use case, and we note that the proposed TRACER is also applicable to other domains, e.g., fintech. We evaluate the accuracy of TRACER extensively in two real-world hospital datasets, and our doctors/clinicians further validate the interpretability of TRACER in both the patient level and the feature level. Besides, TRACER is also validated in a high stakes financial application and a critical temperature forecasting application. The experimental results confirm that TRACER facilitates both accurate and interpretable analytics for high stakes applications.

An unprecedented volume of data is currently being generated across the globe with no less than an estimated 2.5 quintillion (1018) bytes of data each day at our current pace. The variety of formats in which this data is being produced, and its structural complexity, are also on the rise. Collectively, these factors are driving demand among institutions for advanced analytics to generate actionable insights. At the most basic level, machine learning encompasses the use of computational algorithms more advanced than the analytics methods (data mining approaches, for example) traditionally employed to deliver insights into large datasets. Machine learning techniques are firmly rooted in the science of statistics and have valuable applications not least in financial services.

Modern variational inference (VI) uses stochastic gradients to avoid intractable expectations, enabling large-scale probabilistic inference in complex models. VI posits a family of approximating distributions $q$ and then finds the member of that family that is closest to the exact posterior $p$. Traditionally, VI algorithms minimize the "exclusive KL" KL$(q\|p)$, often for computational convenience. Recent research, however, has also focused on the "inclusive KL" KL$(p\|q)$, which has good statistical properties that makes it more appropriate for certain inference problems. This paper develops a simple algorithm for reliably minimizing the inclusive KL. Consider a valid MCMC method, a Markov chain whose stationary distribution is $p$. The algorithm we develop iteratively samples the chain $z[k]$, and then uses those samples to follow the score function of the variational approximation, $\nabla \log q(z[k])$ with a Robbins-Monro step-size schedule. This method, which we call Markovian score climbing (MSC), converges to a local optimum of the inclusive KL. It does not suffer from the systematic errors inherent in existing methods, such as Reweighted Wake-Sleep and Neural Adaptive Sequential Monte Carlo, which lead to bias in their final estimates. In a variant that ties the variational approximation directly to the Markov chain, MSC further provides a new algorithm that melds VI and MCMC. We illustrate convergence on a toy model and demonstrate the utility of MSC on Bayesian probit regression for classification as well as a stochastic volatility model for financial data.

As massive data are produced from small gadgets, federated learning on mobile devices has become an emerging trend. In the federated setting, Stochastic Gradient Descent (SGD) has been widely used in federated learning for various machine learning models. To prevent privacy leakages from gradients that are calculated on users' sensitive data, local differential privacy (LDP) has been considered as a privacy guarantee in federated SGD recently. However, the existing solutions have a dimension dependency problem: the injected noise is substantially proportional to the dimension $d$. In this work, we propose a two-stage framework FedSel for federated SGD under LDP to relieve this problem. Our key idea is that not all dimensions are equally important so that we privately select Top-k dimensions according to their contributions in each iteration of federated SGD. Specifically, we propose three private dimension selection mechanisms and adapt the gradient accumulation technique to stabilize the learning process with noisy updates. We also theoretically analyze privacy, accuracy and time complexity of FedSel, which outperforms the state-of-the-art solutions. Experiments on real-world and synthetic datasets verify the effectiveness and efficiency of our framework.

Research in the supervised learning algorithms field implicitly assumes that training data is labeled by domain experts or at least semi-professional labelers accessible through crowdsourcing services like Amazon Mechanical Turk. With the advent of the Internet, data has become abundant and a large number of machine learning based systems started being trained with user-generated data, using categorical data as true labels. However, little work has been done in the area of supervised learning with user-defined labels where users are not necessarily experts and might be motivated to provide incorrect labels in order to improve their own utility from the system. In this article, we propose two types of classes in user-defined labels: subjective class and objective class - showing that the objective classes are as reliable as if they were provided by domain experts, whereas the subjective classes are subject to bias and manipulation by the user. We define this as a subjective class issue and provide a framework for detecting subjective labels in a dataset without querying oracle. Using this framework, data mining practitioners can detect a subjective class at an early stage of their projects, and avoid wasting their precious time and resources by dealing with subjective class problem with traditional machine learning techniques.

In this paper we present a novel approach to optimise tactical and strategic decision making in football (soccer). We model the game of football as a multi-stage game which is made up from a Bayesian game to model the pre-match decisions and a stochastic game to model the in-match state transitions and decisions. Using this formulation, we propose a method to predict the probability of game outcomes and the payoffs of team actions. Building upon this, we develop algorithms to optimise team formation and in-game tactics with different objectives. Empirical evaluation of our approach on real-world datasets from 760 matches shows that by using optimised tactics from our Bayesian and stochastic games, we can increase a team chances of winning by up to 16.1\% and 3.4\% respectively.

The BOLD signal of resting-state fMRI (rs-fMRI) records the functional brain connectivity in a rich dynamic spatio-temporal setting. However, existing methods applied to rs-fMRI often fail to consider both spatial and temporal characteristics of the data. They either neglect the functional dependency between different brain regions in a network or discard the information in the temporal dynamics of brain activity. To overcome those shortcomings, we propose to formulate functional connectivity networks within the context of spatio-temporal graphs. We then train a spatio-temporal graph convolutional network (ST-GCN) on short sub-sequences of the BOLD time series to model the non-stationary nature of functional connectivity. We simultaneously learn the graph edge importance within ST-GCN to enable interpretation of functional connectivities contributing to the prediction model. In analyzing the rs-fMRI of the Human Connectome Project (HCP, N=1,091) and the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA, N=773), ST-GCN is significantly more accurate than common approaches in predicting gender and age based on BOLD signals. The matrix recording edge importance localizes brain regions and functional connections with significant aging and sex effects, which are verified by the neuroscience literature.

Distributed Stochastic Gradient Descent (SGD) when run in a synchronous manner, suffers from delays in runtime as it waits for the slowest workers (stragglers). Asynchronous methods can alleviate stragglers, but cause gradient staleness that can adversely affect the convergence error. In this work, we present a novel theoretical characterization of the speedup offered by asynchronous methods by analyzing the trade-off between the error in the trained model and the actual training runtime(wallclock time). The main novelty in our work is that our runtime analysis considers random straggling delays, which helps us design and compare distributed SGD algorithms that strike a balance between straggling and staleness. We also provide a new error convergence analysis of asynchronous SGD variants without bounded or exponential delay assumptions. Finally, based on our theoretical characterization of the error-runtime trade-off, we propose a method of gradually varying synchronicity in distributed SGD and demonstrate its performance on CIFAR10 dataset.

In this paper, our focus is the connection and influence of language technologies on the research in neurolinguistics. We present a review of brain imaging-based neurolinguistic studies with a focus on the natural language representations, such as word embeddings and pre-trained language models. Mutual enrichment of neurolinguistics and language technologies leads to development of brain-aware natural language representations. The importance of this research area is emphasized by medical applications.

In the context of neural network models, overparametrization refers to the phenomena whereby these models appear to generalize well on the unseen data, even though the number of parameters significantly exceeds the sample sizes, and the model perfectly fits the in-training data. A conventional explanation of this phenomena is based on self-regularization properties of algorithms used to train the data. In this paper we prove a series of results which provide a somewhat diverging explanation. Adopting a teacher/student model where the teacher network is used to generate the predictions and student network is trained on the observed labeled data, and then tested on out-of-sample data, we show that any student network interpolating the data generated by a teacher network generalizes well, provided that the sample size is at least an explicit quantity controlled by data dimension and approximation guarantee alone, regardless of the number of internal nodes of either teacher or student network. Our claim is based on approximating both teacher and student networks by polynomial (tensor) regression models with degree depending on the desired accuracy and network depth only. Such a parametrization notably does not depend on the number of internal nodes. Thus a message implied by our results is that parametrizing wide neural networks by the number of hidden nodes is misleading, and a more fitting measure of parametrization complexity is the number of regression coefficients associated with tensorized data. In particular, this somewhat reconciles the generalization ability of neural networks with more classical statistical notions of data complexity and generalization bounds. Our empirical results on MNIST and Fashion-MNIST datasets indeed confirm that tensorized regression achieves a good out-of-sample performance, even when the degree of the tensor is at most two.