Whether the visualizations provided by the bias-detection software will be enough to understand and more importantly explain a complex deep learning AI algorithm remains to be seen, as it has so far been impossible to understand how AI systems reach decisions. This is therefore a big claim from IBM, as only last year Tommi Jaakkola, an MIT professor working on applications of Machine Learning, said: 'If you had a very small neural network, you might be able to understand it, but once it becomes very large, and it has thousands of units per layer and maybe hundreds of layers, then it becomes quite un-understandable.' The problem is not only that we can't see what deep learning algorithms are doing, but we also can't understand their workings, so will IBM's visualization software solve anything?

Medical imaging creates tremendous amounts of data: many emergency room radiologists must examine as many as 200 cases each day, and some medical studies contain up to 3,000 images. Each patient's image collection can contain 250GB of data, ultimately creating collections across organizations that are petabytes in size. Within IBM Research, we see potential in applying AI to help radiologists sift through this information, including imaging analysis from breast, liver, and lung exams. IBM researchers are applying deep learning to discover ways to overcome some of the technical challenges that AI can face when analyzing X-rays and other medical images. Their latest findings will be presented at the 21st International Conference on Medical Image Computing & Computer Assisted Intervention in Granada, Spain, from September 16 to 20.

The Industrial Internet of Things drastically increases connectivity of devices in industrial applications. In addition to the benefits in efficiency, scalability and ease of use, this creates novel attack surfaces. Historically, industrial networks and protocols do not contain means of security, such as authentication and encryption, that are made necessary by this development. Thus, industrial IT-security is needed. In this work, emulated industrial network data is transformed into a time series and analysed with three different algorithms. The data contains labeled attacks, so the performance can be evaluated. Matrix Profiles perform well with almost no parameterisation needed. Seasonal Autoregressive Integrated Moving Average performs well in the presence of noise, requiring parameterisation effort. Long Short Term Memory-based neural networks perform mediocre while requiring a high training- and parameterisation effort.

The goal behind Domain Adaptation (DA) is to leverage the labeled examples from a source domain so as to infer an accurate model in a target domain where labels are not available or in scarce at the best. A state-of-the-art approach for the DA is due to (Ganin et al. 2016), known as DANN, where they attempt to induce a common representation of source and target domains via adversarial training. This approach requires a large number of labeled examples from the source domain to be able to infer a good model for the target domain. However, in many situations obtaining labels in the source domain is expensive which results in deteriorated performance of DANN and limits its applicability in such scenarios. In this paper, we propose a novel approach to overcome this limitation. In our work, we first establish that DANN reduces the original DA problem into a semi-supervised learning problem over the space of common representation. Next, we propose a learning approach, namely TransDANN, that amalgamates adversarial learning and transductive learning to mitigate the detrimental impact of limited source labels and yields improved performance. Experimental results (both on text and images) show a significant boost in the performance of TransDANN over DANN under such scenarios. We also provide theoretical justification for the performance boost.

Community detection is a challenging, yet crucial, problem while mining large-scale graph structured data. Most existing approaches solve this problem by mapping nodes into a vector space and performing unsupervised learning with the resulting embeddings. In cases where multiple types of connectivity patterns exist for the set of nodes, commonly modeled as multilayer graphs, new strategies are required to model the inter-layer dependencies in order to perform effective inferencing. In this paper, we focus on learning embeddings for each node of a multilayer graph through neural modeling techniques, such that the complex dependencies can be concisely encoded into low-dimensional representations. Referred to as multilayer graph embeddings, these representations can be utilized for discovering community structure in a scalable fashion, even with a large number of layers. Furthermore, in order to ensure that the semantics that persist over a longer range in the network are well modeled, we propose to refine the multilayer embeddings via a proxy clustering loss and a graph modularity measure. Using real-world datasets, we demonstrate that this algorithm generates scalable and robust representations, and outperforms existing multilayer community detection approaches.

In this paper we present an end-to-end deep learning framework to turn images that show dynamic content, such as vehicles or pedestrians, into realistic static frames. This objective encounters two main challenges: detecting the dynamic objects, and inpainting the static occluded background. The second challenge is approached with a conditional generative adversarial model that, taking as input the original dynamic image and the computed dynamic/static binary mask, is capable of generating the final static image. The former challenge is addressed by the use of a convolutional network that learns a multi-class semantic segmentation of the image. The objective of this network is producing an accurate segmentation and helping the previous generative model to output a realistic static image. These generated images can be used for applications such as virtual reality or vision-based robot localization purposes. To validate our approach, we show both qualitative and quantitative comparisons against other inpainting methods by removing the dynamic objects and hallucinating the static structure behind them. Furthermore, to demonstrate the potential of our results, we conduct pilot experiments showing the benefits of our proposal for visual place recognition. Code has been made available on https://github.com/bertabescos/EmptyCities.

NORD (Neural Operations Research & Development) is an open source distributed deep learning architectural research framework, based on PyTorch, MPI and Horovod. It aims to make research of deep architectures easier for experts of different domains, in order to accelerate the process of finding better architectures, as well as study the best architectures generated for different datasets. Although currently under heavy development, the framework aims to allow the easy implementation of different design and optimization method families (optimization algorithms, meta-heuristics, reinforcement learning etc.) as well as the fair comparison between them. Furthermore, due to the computational resources required in order to optimize and evaluate network architectures, it leverage the use of distributed computing, while aiming to minimize the researcher's overhead required to implement it. Moreover, it strives to make the creation of architectures more intuitive, by implementing network descriptors, allowing to separately define the architecture's nodes and connections. In this paper, we present the framework's current state of development, while presenting its basic concepts, providing simple examples as well as their experimental results.

Quizlet is the most popular online learning tool in the United States, and is used by over 2/3 of high school students, and 1/2 of college students. With more than 95% of Quizlet users reporting improved grades as a result, the platform has become the de-facto tool used in millions of classrooms. In this paper, we explore the task of recommending suitable content for a student to study, given their prior interests, as well as what their peers are studying. We propose a novel approach, i.e. Neural Educational Recommendation Engine (NERE), to recommend educational content by leveraging student behaviors rather than ratings. We have found that this approach better captures social factors that are more aligned with learning. NERE is based on a recurrent neural network that includes collaborative and content-based approaches for recommendation, and takes into account any particular student's speed, mastery, and experience to recommend the appropriate task. We train NERE by jointly learning the user embeddings and content embeddings, and attempt to predict the content embedding for the final timestamp. We also develop a confidence estimator for our neural network, which is a crucial requirement for productionizing this model. We apply NERE to Quizlet's proprietary dataset, and present our results. We achieved an R^2 score of 0.81 in the content embedding space, and a recall score of 54% on our 100 nearest neighbors. This vastly exceeds the recall@100 score of 12% that a standard matrix-factorization approach provides. We conclude with a discussion on how NERE will be deployed, and position our work as one of the first educational recommender systems for the K-12 space.

Automated scoring engines are increasingly being used to score the free-form text responses that students give to questions. Such engines are not designed to appropriately deal with responses that a human reader would find alarming such as those that indicate an intention to self-harm or harm others, responses that allude to drug abuse or sexual abuse or any response that would elicit concern for the student writing the response. Our neural network models have been designed to help identify these anomalous responses from a large collection of typical responses that students give. The responses identified by the neural network can be assessed for urgency, severity, and validity more quickly by a team of reviewers than otherwise possible. Given the anomalous nature of these types of responses, our goal is to maximize the chance of flagging these responses for review given the constraint that only a fixed percentage of responses can viably be assessed by a team of reviewers.

Obesity is a serious public health concern world-wide, which increases the risk of many diseases, including hypertension, stroke, and type 2 diabetes. To tackle this problem, researchers across the health ecosystem are collecting diverse types of data, which includes biomedical, behavioral and activity, and utilizing machine learning techniques to mine hidden patterns for obesity status improvement prediction. While existing machine learning methods such as Recurrent Neural Networks (RNNs) can provide exceptional results, it is challenging to discover hidden patterns of the sequential data due to the irregular observation time instances. Meanwhile, the lack of understanding of why those learning models are effective also limits further improvements on their architectures. Thus, in this work, we develop a RNN based time-aware architecture to tackle the challenging problem of handling irregular observation times and relevant feature extractions from longitudinal patient records for obesity status improvement prediction. To improve the prediction performance, we train our model using two data sources: (i) electronic medical records containing information regarding lab tests, diagnoses, and demographics; (ii) continuous activity data collected from popular wearables. Evaluations of real-world data demonstrate that our proposed method can capture the underlying structures in users' time sequences with irregularities, and achieve an accuracy of 77-86% in predicting the obesity status improvement.