Rachel Thomas is a deep learning researcher, she is the cofounder of fats.ai which created the "Practical Deep Learning for Coders" course that over 100,000 students have taken. In her talk she explain why Artificial Intelligence should be accessible to all of us, even without a math background. Rachel was selected by Forbes as one of 20 Incredible Women in AI, earned her math PhD at Duke, and was an early engineer at Uber. Her writing about AI and has been read by over half a million people and has been translated into Chinese, Spanish, Korean, & Portuguese. Co-Founder fast.ai and Researcher at USF. Mathematician, Deep Learning researcher, bringing and teaching Deep Learning to the mainstream.

In natural hazard warning systems fast decision making is vital to avoid catastrophes. Decision making at the edge of a wireless sensor network promises fast response times but is limited by the availability of energy, data transfer speed, processing and memory constraints. In this work we present a realization of a wireless sensor network for hazard monitoring which is based on an array of event-triggered seismic sensors with advanced signal processing and characterization capabilities for a novel co-detection technique. On the one hand we leverage an ultra-low power, threshold-triggering circuit paired with on-demand digital signal acquisition capable of extracting relevant information exactly when it matters most and not wasting precious resources when nothing can be observed. On the other hand we use machine-learning-based classification implemented on low-power, off-the-shelf microcontrollers to avoid false positive warnings and to actively identify humans in hazard zones. The sensors' response time and memory requirement is substantially improved by pipelining the inference of a convolutional neural network. In this way, convolutional neural networks that would not run unmodified on a memory constrained device can be executed in real-time and at scale on low-power embedded devices.

Deep neural networks (DNNs) are computationally/memory-intensive and vulnerable to adversarial attacks, making them prohibitive in some real-world applications. By converting dense models into sparse ones, pruning appears to be a promising solution to reducing the computation/memory cost. This paper studies classification models, especially DNN-based ones, to demonstrate that there exists intrinsic relationships between their sparsity and adversarial robustness. Our analyses reveal, both theoretically and empirically, that nonlinear DNN-based classifiers behave differently under $l_2$ attacks from some linear ones. We further demonstrate that an appropriately higher model sparsity implies better robustness of nonlinear DNNs, whereas over-sparsified models can be more difficult to resist adversarial examples.

Solving partial differential equations (PDEs) underlies much of applied mathematics and engineering, ranging from computer graphics and financial pricing, to civil engineering and weather prediction. Conventional approaches to prediction in PDE models rely on numerical solvers and require substantial computing resources in the model-application phase. While in some application domains, such as structural engineering, the longer run-times may be acceptable, in domains with rapid decay of value of the prediction, such as weather forecasting, the run-time of the solver is of paramount importance. In many such applications, the ability to generate large volumes of data facilitates the use of surrogate or reduced-order models [1] obtained using deep artificial neural networks [2]. Although the observation that artificial neural networks could be applied to physical models is not new [3, 4, 5, 6, 7, 8, 9, 5, 10], and indeed, it is seen as one of the key trends [11, 12, 13] on the interface of applied mathematics, data science, and deep learning, their applications did not reach the level of success observed in the field of the image classification, speech recognition, machine translation, and other problems processing unstructured high-dimensional data, yet. A key issue faced by applications of deep-learning techniques to physical models is their scalability. Even very recent research on deep-learning for physical models [14, 15, 16] uses a solver for PDEs to obtain hundreds of thousands of outputs. The deep learning can then be seen as means of nonlinear regression between the inputs and outputs.

The recent development of deep learning has led to dramatic progress in multiple research fields, and this technique has naturally found applications in autonomous vehicles. The use of deep learning to perform perceptive tasks such as image segmentation has been widely researched in the last few years, and highly efficient neural network architectures are now available for such tasks. More recently, several teams have proposed taking deep learning a step further, by training so-called "end-to-end" algorithms to directly output vehicle controls from raw sensor data (see, in particular, the seminal work in [1]). Although end-to-end driving is highly appealing, as it removes the need to design motion planning and control algorithms by hand, handing the safety of the car occupants to a software operating as a black box seems problematic. A possible workaround to this downside is to use "forensics" techniques that can, to a certain extent, help understand the behavior of deep neural networks [2]. We choose a different approach consisting in breaking down complexity by training simpler, mono-task neural networks to solve specific problems arising in autonomous driving; we argue that the reduced complexity of individual tasks allows much easier testing and validation. In this article, we focus on the problem of controlling a car-like vehicle in highly dynamic situations, for instance to perform evasive maneuvers in face of an obstacle. A particular challenge in such scenarios is the important coupling between longitudinal and lateral dynamics when nearing the vehicle's handling limits, which requires highly detailed Published in the IEEE 2018 International Conference on Intelligent Transportation Systems (ITSC 2018). This work was supported by the international Chair MINES ParisTech - Peugeot-Citro en - Safran - V aleo on ground vehicle automation.

Tuning machine learning models, particularly deep learning architectures, is a complex process. Automated hyperparameter tuning algorithms often depend on specific optimization metrics. However, in many situations, a developer trades one metric against another: accuracy versus overfitting, precision versus recall, smaller models and accuracy, etc. With deep learning, not only are the model's representations opaque, the model's behavior when parameters "knobs" are changed may also be unpredictable. Thus, picking the "best" model often requires time-consuming model comparison. In this work, we introduce LAMVI-2, a visual analytics system to support a developer in comparing hyperparameter settings and outcomes. By focusing on word-embedding models ("deep learning for text") we integrate views to compare both high-level statistics as well as internal model behaviors (e.g., comparing word 'distances'). We demonstrate how developers can work with LAMVI-2 to more quickly and accurately narrow down an appropriate and effective application-specific model.

Deep learning models exhibit state-of-the-art performance for many predictive healthcare tasks using electronic health records (EHR) data, but these models typically require training data volume that exceeds the capacity of most healthcare systems. External resources such as medical ontologies are used to bridge the data volume constraint, but this approach is often not directly applicable or useful because of inconsistencies with terminology. To solve the data insufficiency challenge, we leverage the inherent multilevel structure of EHR data and, in particular, the encoded relationships among medical codes. We propose Multilevel Medical Embedding (MiME) which learns the multilevel embedding of EHR data while jointly performing auxiliary prediction tasks that rely on this inherent EHR structure without the need for external labels. We conducted two prediction tasks, heart failure prediction and sequential disease prediction, where MiME outperformed baseline methods in diverse evaluation settings. In particular, MiME consistently outperformed all baselines when predicting heart failure on datasets of different volumes, especially demonstrating the greatest performance improvement (15% relative gain in PR-AUC over the best baseline) on the smallest dataset, demonstrating its ability to effectively model the multilevel structure of EHR data.

The field of few-shot learning has recently seen substantial advancements. Most of these advancements came from casting few-shot learning as a meta-learning problem. Model Agnostic Meta Learning or MAML is currently one of the best approaches for few-shot learning via meta-learning. MAML is simple, elegant and very powerful, however, it has a variety of issues, such as being very sensitive to neural network architectures, often leading to instability during training, requiring arduous hyperparameter searches to stabilize training and achieve high generalization and being very computationally expensive at both training and inference times. In this paper, we propose various modifications to MAML that not only stabilize the system, but also substantially improve the generalization performance, convergence speed and computational overhead of MAML, which we call MAML . The human capacity to learn new concepts using only a handful of samples is immense. In stark contrast, modern deep neural networks need, at a minimum, thousands of samples before they begin to learn representations that can generalize well to unseen data-points (Krizhevsky et al., 2012; Huang et al., 2017), and mostly fail when the data available is scarce. The fact that standard deep neural networks fail in the small data regime can provide hints about some of their potential shortcomings. Solving those shortcomings has the potential to open the door to understanding intelligence and advancing Artificial Intelligence. Few-shot learning encapsulates a family of methods that can learn new concepts with only a handful of data-points (usually 1-5 samples per concept).

Adversarial machine learning is a fast growing research area, which considers the scenarios when machine learning systems may face potential adversarial attackers, who intentionally synthesize input data to make a well-trained model to make mistake. It always involves a defending side, usually a classifier, and an attacking side that aims to cause incorrect output. The earliest studies on the adversarial examples for machine learning algorithms start from the information security area, which considers a much wider varieties of attacking methods. But recent research focus that popularized by the deep learning community places strong emphasis on how the "imperceivable" perturbations on the normal inputs may cause dramatic mistakes by the deep learning with supposed super-human accuracy. This paper serves to give a comprehensive introduction to a range of aspects of the adversarial deep learning topic, including its foundations, typical attacking and defending strategies, and some extended studies.

Deep learning models for semantic segmentation of images require large amounts of data. In the medical imaging domain, acquiring sufficient data is a significant challenge. Labeling medical image data requires expert knowledge. Collaboration between institutions could address this challenge, but sharing medical data to a centralized location faces various legal, privacy, technical, and data-ownership challenges, especially among international institutions. In this study, we introduce the first use of federated learning for multi-institutional collaboration, enabling deep learning modeling without sharing patient data. Our quantitative results demonstrate that the performance of federated semantic segmentation models (Dice 0.852) on multimodal brain scans is similar to that of models trained by sharing data (Dice 0.862). We compare federated learning with two alternative collaborative learning methods and find that they fail to match the performance of federated learning.