We can add suggesting and proving mathematical theorems to the long list of what artificial intelligence is capable of: Mathematicians and AI experts have teamed up to demonstrate how machine learning can open up new avenues to explore in the field. While mathematicians have been using computers to discover patterns for decades, the increasing power of machine learning means that these networks can work through huge swathes of data and identify patterns that haven't been spotted before. In a newly published study, a research team used artificial intelligence systems developed by DeepMind, the same company that has been deploying AI to solve tricky biology problems and improve the accuracy of weather forecasts, to unknot some long-standing math problems. "Problems in mathematics are widely regarded as some of the most intellectually challenging problems out there," says mathematician Geordie Williamson from the University of Sydney in Australia. "While mathematicians have used machine learning to assist in the analysis of complex data sets, this is the first time we have used computers to help us formulate conjectures or suggest possible lines of attack for unproven ideas in mathematics."

The increasing availability of large collections of electronic health record (EHR) data and unprecedented technical advances in deep learning (DL) have sparked a surge of research interest in developing DL based clinical decision support systems for diagnosis, prognosis, and treatment. Despite the recognition of the value of deep learning in healthcare, impediments to further adoption in real healthcare settings remain due to the black-box nature of DL. Therefore, there is an emerging need for interpretable DL, which allows end users to evaluate the model decision making to know whether to accept or reject predictions and recommendations before an action is taken. In this review, we focus on the interpretability of the DL models in healthcare. We start by introducing the methods for interpretability in depth and comprehensively as a methodological reference for future researchers or clinical practitioners in this field. Besides the methods' details, we also include a discussion of advantages and disadvantages of these methods and which scenarios each of them is suitable for, so that interested readers can know how to compare and choose among them for use. Moreover, we discuss how these methods, originally developed for solving general-domain problems, have been adapted and applied to healthcare problems and how they can help physicians better understand these data-driven technologies. Overall, we hope this survey can help researchers and practitioners in both artificial intelligence (AI) and clinical fields understand what methods we have for enhancing the interpretability of their DL models and choose the optimal one accordingly.

Continual learning (CL) learns a sequence of tasks incrementally with the goal of achieving two main objectives: overcoming catastrophic forgetting (CF) and encouraging knowledge transfer (KT) across tasks. However, most existing techniques focus only on overcoming CF and have no mechanism to encourage KT, and thus do not do well in KT. Although several papers have tried to deal with both CF and KT, our experiments show that they suffer from serious CF when the tasks do not have much shared knowledge. Another observation is that most current CL methods do not use pre-trained models, but it has been shown that such models can significantly improve the end task performance. For example, in natural language processing, fine-tuning a BERT-like pre-trained language model is one of the most effective approaches. However, for CL, this approach suffers from serious CF. An interesting question is how to make the best use of pre-trained models for CL. This paper proposes a novel model called CTR to solve these problems. Our experimental results demonstrate the effectiveness of CTR.

Mobile devices are indispensable sources of big data. Federated learning (FL) has a great potential in exploiting these private data by exchanging locally trained models instead of their raw data. However, mobile devices are often energy limited and wirelessly connected, and FL cannot cope flexibly with their heterogeneous and time-varying energy capacity and communication throughput, limiting the adoption. Motivated by these issues, we propose a novel energy and communication efficient FL framework, coined SlimFL. To resolve the heterogeneous energy capacity problem, each device in SlimFL runs a width-adjustable slimmable neural network (SNN). To address the heterogeneous communication throughput problem, each full-width (1.0x) SNN model and its half-width ($0.5$x) model are superposition-coded before transmission, and successively decoded after reception as the 0.5x or $1.0$x model depending on the channel quality. Simulation results show that SlimFL can simultaneously train both $0.5$x and $1.0$x models with reasonable accuracy and convergence speed, compared to its vanilla FL counterpart separately training the two models using $2$x more communication resources. Surprisingly, SlimFL achieves even higher accuracy with lower energy footprints than vanilla FL for poor channels and non-IID data distributions, under which vanilla FL converges slowly.

This paper introduces the proposed automatic minuting system of the Hitachi team for the First Shared Task on Automatic Minuting (AutoMin-2021). We utilize a reference-free approach (i.e., without using training minutes) for automatic minuting (Task A), which first splits a transcript into blocks on the basis of topics and subsequently summarizes those blocks with a pre-trained BART model fine-tuned on a summarization corpus of chat dialogue. In addition, we apply a technique of argument mining to the generated minutes, reorganizing them in a well-structured and coherent way. We utilize multiple relevance scores to determine whether or not a minute is derived from the same meeting when either a transcript or another minute is given (Task B and C). On top of those scores, we train a conventional machine learning model to bind them and to make final decisions. Consequently, our approach for Task A achieve the best adequacy score among all submissions and close performance to the best system in terms of grammatical correctness and fluency. For Task B and C, the proposed model successfully outperformed a majority vote baseline.

Data augmentation is an important component in the robustness evaluation of models in natural language processing (NLP) and in enhancing the diversity of the data they are trained on. In this paper, we present NL-Augmenter, a new participatory Python-based natural language augmentation framework which supports the creation of both transformations (modifications to the data) and filters (data splits according to specific features). We describe the framework and an initial set of 117 transformations and 23 filters for a variety of natural language tasks. We demonstrate the efficacy of NL-Augmenter by using several of its transformations to analyze the robustness of popular natural language models. The infrastructure, datacards and robustness analysis results are available publicly on the NL-Augmenter repository (\url{https://github.com/GEM-benchmark/NL-Augmenter}).

Social media, professional sports, and video games are driving rapid growth in live video streaming, on platforms such as Twitch and YouTube Live. Live streaming experience is very susceptible to short-time-scale network congestion since client playback buffers are often no more than a few seconds. Unfortunately, identifying such streams and measuring their QoE for network management is challenging, since content providers largely use the same delivery infrastructure for live and video-on-demand (VoD) streaming, and packet inspection techniques (including SNI/DNS query monitoring) cannot always distinguish between the two. In this paper, we design, build, and deploy ReCLive: a machine learning method for live video detection and QoE measurement based on network-level behavioral characteristics. Our contributions are four-fold: (1) We analyze about 23,000 video streams from Twitch and YouTube, and identify key features in their traffic profile that differentiate live and on-demand streaming. We release our traffic traces as open data to the public; (2) We develop an LSTM-based binary classifier model that distinguishes live from on-demand streams in real-time with over 95% accuracy across providers; (3) We develop a method that estimates QoE metrics of live streaming flows in terms of resolution and buffer stall events with overall accuracies of 93% and 90%, respectively; and (4) Finally, we prototype our solution, train it in the lab, and deploy it in a live ISP network serving more than 7,000 subscribers. Our method provides ISPs with fine-grained visibility into live video streams, enabling them to measure and improve user experience.

Data-driven forecasts of air quality have recently achieved more accurate short-term predictions. Despite their success, most of the current data-driven solutions lack proper quantifications of model uncertainty that communicate how much to trust the forecasts. Recently, several practical tools to estimate uncertainty have been developed in probabilistic deep learning. However, there have not been empirical applications and extensive comparisons of these tools in the domain of air quality forecasts. Therefore, this work applies state-of-the-art techniques of uncertainty quantification in a real-world setting of air quality forecasts. Through extensive experiments, we describe training probabilistic models and evaluate their predictive uncertainties based on empirical performance, reliability of confidence estimate, and practical applicability. We also propose improving these models using "free" adversarial training and exploiting temporal and spatial correlation inherent in air quality data. Our experiments demonstrate that the proposed models perform better than previous works in quantifying uncertainty in data-driven air quality forecasts. Overall, Bayesian neural networks provide a more reliable uncertainty estimate but can be challenging to implement and scale. Other scalable methods, such as deep ensemble, Monte Carlo (MC) dropout, and stochastic weight averaging-Gaussian (SWAG), can perform well if applied correctly but with different tradeoffs and slight variations in performance metrics. Finally, our results show the practical impact of uncertainty estimation and demonstrate that, indeed, probabilistic models are more suitable for making informed decisions. Code and dataset are available at \url{https://github.com/Abdulmajid-Murad/deep_probabilistic_forecast}

Active learning is an established technique to reduce the labeling cost to build high-quality machine learning models. A core component of active learning is the acquisition function that determines which data should be selected to annotate. State-of-the-art acquisition functions -- and more largely, active learning techniques -- have been designed to maximize the clean performance (e.g. accuracy) and have disregarded robustness, an important quality property that has received increasing attention. Active learning, therefore, produces models that are accurate but not robust. In this paper, we propose \emph{robust active learning}, an active learning process that integrates adversarial training -- the most established method to produce robust models. Via an empirical study on 11 acquisition functions, 4 datasets, 6 DNN architectures, and 15105 trained DNNs, we show that robust active learning can produce models with the robustness (accuracy on adversarial examples) ranging from 2.35\% to 63.85\%, whereas standard active learning systematically achieves negligible robustness (less than 0.20\%). Our study also reveals, however, that the acquisition functions that perform well on accuracy are worse than random sampling when it comes to robustness. We, therefore, examine the reasons behind this and devise a new acquisition function that targets both clean performance and robustness. Our acquisition function -- named density-based robust sampling with entropy (DRE) -- outperforms the other acquisition functions (including random) in terms of robustness by up to 24.40\% (3.84\% than random particularly), while remaining competitive on accuracy. Additionally, we prove that DRE is applicable as a test selection metric for model retraining and stands out from all compared functions by up to 8.21\% robustness.

Intention recognition is an important step to facilitate collaboration in multi-agent systems. Existing work mainly focuses on intention recognition in a single-agent setting and uses a descriptive model, e.g. Bayesian networks, in the recognition process. In this paper, we resort to a prescriptive approach to model agents' behaviour where which their intentions are hidden in implementing their plans. We introduce landmarks into the behavioural model therefore enhancing informative features for identifying common intentions for multiple agents. We further refine the model by focusing only action sequences in their plan and provide a light model for identifying and comparing their intentions. The new models provide a simple approach of grouping agents' common intentions upon partial plans observed in agents' interactions. We provide experimental results in support.