We trained and evaluated a localization-based deep CNN for breast cancer screening exam classification on over 200,000 exams (over 1,000,000 images). Our model achieves an AUC of 0.919 in predicting malignancy in patients undergoing breast cancer screening, reducing the error rate of the baseline (Wu et al., 2019a) by 23%. In addition, the models generates bounding boxes for benign and malignant findings, providing interpretable predictions.

Many common sequential data sources, such as source code and natural language, have a natural tree-structured representation. These trees can be generated by fitting a sequence to a grammar, yielding a hierarchical ordering of the tokens in the sequence. This structure encodes a high degree of syntactic information, making it ideal for problems such as grammar correction. However, little work has been done to develop neural networks that can operate on and exploit tree-structured data. In this paper we present the Tree-Transformer \textemdash{} a novel neural network architecture designed to translate between arbitrary input and output trees. We applied this architecture to correction tasks in both the source code and natural language domains. On source code, our model achieved an improvement of $25\%$ $\text{F}0.5$ over the best sequential method. On natural language, we achieved comparable results to the most complex state of the art systems, obtaining a $10\%$ improvement in recall on the CoNLL 2014 benchmark and the highest to date $\text{F}0.5$ score on the AESW benchmark of $50.43$.

While the trajectory without the vehicle model appears reasonable, it is physically impossible for a two-axle vehicle to execute its motion in such manner because its rear wheels cannot turn. The proposed approach outputs a trajectory that is kinematically feasible and correctly predicts that the actor will encroach into the neighboring lane. We summarize the main contributions of our work below: - We combine powerful deep methods with a kinematic two-axle vehicle motion model in order to output trajectory predictions with guaranteed physical realism; - While the idea is general and applicable to any deep architecture, we present an example application to a recently proposed state-of-the-art motion prediction method, using raster-ized images of vehicle context as input to convolutional neural networks (CNNs) [7]; - We evaluate the method on a large-scale, real-world data set collected by a fleet of SDVs, showing that the system provides accurate, kinematically feasible predictions that outperform the existing state-of-the-art. 2 Related work 2.1 Motion prediction in autonomous driving Accurate motion prediction of other vehicles is a critical component in many autonomous driving systems [9, 10, 11]. Prediction provides an estimate of future world state, which can be used to plan an optimal path for the SDV through a dynamic traffic environment. The current state (e.g., position, speed, acceleration) of vehicles around a SDV can be estimated using techniques such as a Kalman filter (KF) [12, 13]. A common approach for short time horizon predictions of future motion is to assume that the driver will not change any control inputs (steering, accelerator) and simply propagate vehicle's current estimated state over time using a physical model (e.g., a vehicle motion model) that captures the underlying kinematics [9]. For longer time horizons the performance of this approach degrades as the underlying assumption of constant controls becomes increasingly unlikely.

Various domain users are increasingly leveraging real-time social media data to gain rapid situational awareness. However, due to the high noise in the deluge of data, effectively determining semantically relevant information can be difficult, further complicated by the changing definition of relevancy by each end user for different events. The majority of existing methods for short text relevance classification fail to incorporate users' knowledge into the classification process. Existing methods that incorporate interactive user feedback focus on historical datasets. Therefore, classifiers cannot be interactively retrained for specific events or user-dependent needs in real-time. This limits real-time situational awareness, as streaming data that is incorrectly classified cannot be corrected immediately, permitting the possibility for important incoming data to be incorrectly classified as well. We present a novel interactive learning framework to improve the classification process in which the user iteratively corrects the relevancy of tweets in real-time to train the classification model on-the-fly for immediate predictive improvements. We computationally evaluate our classification model adapted to learn at interactive rates. Our results show that our approach outperforms state-of-the-art machine learning models. In addition, we integrate our framework with the extended Social Media Analytics and Reporting Toolkit (SMART) 2.0 system, allowing the use of our interactive learning framework within a visual analytics system tailored for real-time situational awareness. To demonstrate our framework's effectiveness, we provide domain expert feedback from first responders who used the extended SMART 2.0 system.

Aleatoric uncertainty is an intrinsic property of ill-posed inverse and imaging problems. Its quantification is vital for assessing the reliability of relevant point estimates. In this paper, we propose an efficient framework for quantifying aleatoric uncertainty for deep residual learning and showcase its significant potential on image restoration. In the framework, we divide the conditional probability modeling for the residual variable into a deterministic homo-dimensional level, a stochastic low-dimensional level and a merging level. The low-dimensionality is especially suitable for sparse correlation between image pixels, enables efficient sampling for high dimensional problems and acts as a regularizer for the distribution. Preliminary numerical experiments show that the proposed method can give not only state-of-the-art point estimates of image restoration but also useful associated uncertainty information.

When training clinical prediction models from electronic health records (EHRs), a key concern should be a model's ability to sustain performance over time when deployed, even as care practices, database systems, and population demographics evolve. Due to de-identification requirements, however, current experimental practices for public EHR benchmarks (such as the MIMIC-III critical care dataset) are time agnostic, assigning care records to train or test sets without regard for the actual dates of care. As a result, current benchmarks cannot assess how well models trained on one year generalise to another. In this work, we obtain a Limited Data Use Agreement to access year of care for each record in MIMIC and show that all tested state-of-the-art models decay in prediction quality when trained on historical data and tested on future data, particularly in response to a system-wide record-keeping change in 2008 (0.29 drop in AUROC for mortality prediction, 0.10 drop in AUROC for length-of-stay prediction with a random forest classifier). We further develop a simple yet effective mitigation strategy: by aggregating raw features into expert-defined clinical concepts, we see only a 0.06 drop in AUROC for mortality prediction and a 0.03 drop in AUROC for length-of-stay prediction. We demonstrate that this aggregation strategy outperforms other automatic feature preprocessing techniques aimed at increasing robustness to data drift. We release our aggregated representations and code to encourage more deployable clinical prediction models.

With higher-order neighborhood information of graph network, the accuracy of graph representation learning classification can be significantly improved. However, the current higher order graph convolutional network has a large number of parameters and high computational complexity. Therefore, we propose a Hybrid Lower order and Higher order Graph convolutional networks (HLHG) learning model, which uses weight sharing mechanism to reduce the number of network parameters. To reduce computational complexity, we propose a novel fusion pooling layer to combine the neighborhood information of high order and low order. Theoretically, we compare the model complexity of the proposed model with the other state-of-the-art model. Experimentally, we verify the proposed model on the large-scale text network datasets by supervised learning, and on the citation network datasets by semi-supervised learning. The experimental results show that the proposed model achieves highest classification accuracy with a small set of trainable weight parameters.

Most speech recognition tasks pertain to mapping words across two modalities: acoustic and orthographic. In this work, we suggest learning encoders that map variable-length, acoustic or phonetic, sequences that represent words into fixed-dimensional vectors in a shared latent space; such that the distance between two word vectors represents how closely the two words sound. Instead of directly learning the distances between word vectors, we employ weak supervision and model a binary classification task to predict whether two inputs, one of each modality, represent the same word given a distance threshold. We explore various deep-learning models, bimodal contrastive losses, and techniques for mining hard negative examples such as the semi-supervised technique of self-labeling. Our best model achieves an $F_1$ score of 0.95 for the binary classification task.

Software frameworks for neural networks play a key role in the development and application of deep learning methods. In this paper, we introduce the Chainer framework, which intends to provide a flexible, intuitive, and high performance means of implementing the full range of deep learning models needed by researchers and practitioners. Chainer provides acceleration using Graphics Processing Units with a familiar NumPy-like API through CuPy, supports general and dynamic models in Python through Define-by-Run, and also provides add-on packages for state-of-the-art computer vision models as well as distributed training.

Artificial intelligence trained on health records can now detect kidney injury up to two days before it occurs. The idea is that an advance warning could help doctors intervene earlier to prevent irreversible damage to the kidneys. AIs are already touted as rivals to doctors when it comes to detecting medical conditions such as certain cancers and childhood illnesses. But few undergo rigorous clinical trials, so it's still too early to know whether they are effective in practice. Nenad Tomašev at DeepMind and colleagues trained an algorithm to predict the likelihood that a person who was admitted to hospital would go on to develop acute kidney injury (AKI). They trained the AI using de-identified electronic health records from 703,782 US veterans aged between 18 and 90, who were admitted to hospital between October 2011 and September 2015.