We know content should be valuable, comprehensive, new, relevant, and accurate. Even more fundamentally important than all of those things is that it needs to be authentic. People trust factual information presented with sincere intentions. This era of "fake news" has ushered in a lot of fear for this very reason; we have had to fight to build the authority of our pages and domains to signal that we are worthy of trust. But our industry has yet to face its biggest challenge.

We could not have written this Best of AI without mentioning the 7th edition of the International Conference on Learning Representation (ICLR) which took place in New Orleans at the beginning of the month. Did not get the chance to go to this top-notch research event? We came across a summary of the conference by a researcher from NVIDIA labs. This technical article will give you a great overview of some of the hottest keywords in AI research. Look them up if you do not get them all!

Academics and researchers typically accuse the media of scaremongering and painting dystopian scenarios, especially when it comes to the coverage of Artificial Intelligence (AI)-powered algorithms. They do have a point, especially as machine learning and deep learning algorithms are exactly what power most of the software and smart devices that we use in our daily lives, be they smartphones, cameras, Internet of Things (IoT) devices or voice assistants. Smartphone penetration and advances in image recognition, for instance, are turning phones into powerful at-home diagnostic tools, while these cutting-edge algorithms are helping doctors, researchers and technology companies revolutionize healthcare. Besides, AI in conjunction with IoT (sensors and wearables), robotics, virtual reality (VR) and augmented reality (AR) is playing a very important role. Further, researchers use AI systems to help radiologists improve their ability to diagnose and track prostate cancer. Nvidia researchers have generated synthetic brain MRI images for AI research with the aim of helping doctors learn more about rare brain tumours.

In this paper, we evaluate the PredNet \cite{lotter16} on the Something-something action data set \cite{farzaneh18} and implement the PredNet+, which we train in a multi-task fashion to output both classification labels and predictions. Our idea is to condition video prediction and action classification on each other. We discuss a series of observations about the PredNet and conclude that it does not completely follow the principles of the predictive coding framework.

With the unprecedented success of deep convolutional neural networks came the quest for training always deeper networks. However, while deeper neural networks give better performance when trained appropriately, that depth also translates in memory and computation heavy models, typically with tens of millions of parameters. Several methods have been proposed to leverage redundancies in the network to alleviate this complexity. Either a pretrained network is compressed, e.g. using a low-rank tensor decomposition, or the architecture of the network is directly modified to be more effective. In this paper, we study both approaches in a unified framework, under the lens of tensor decompositions. We show how tensor decomposition applied to the convolutional kernel relates to efficient architectures such as MobileNet. Moreover, we propose a tensor-based method for efficient higher order convolutions, which can be used as a plugin replacement for N-dimensional convolutions. We demonstrate their advantageous properties both theoretically and empirically for image classification, for both 2D and 3D convolutional networks.

Efforts to understand how and why deep learning works have led to a focus on the optimization landscape of training loss. Since optimization to near-zero training loss occurs for many choices of random initialization, it is clear that the landscape contains many global optima (or near-optima). However, the loss can become quite high when interpolating between found optima, suggesting that these optima occur at the bottom of "valleys" surrounded on all sides by high walls. Therefore the phenomenon of mode connectivity (Garipov et al., 2018; Draxler et al., 2018) came as a surprise: optima (at least the ones discovered by gradient-based optimization) are connected by simple paths in the parameter space, on which the loss function is almost constant. In other words, the optima are not walled off in separate valleys as hitherto believed.

In this paper we propose a method of obtaining points of extreme overfitting - parameters of modern neural networks, at which they demonstrate close to 100 % training accuracy, simultaneously with almost zero accuracy on the test sample. Despite the widespread opinion that the overwhelming majority of critical points of the loss function of a neural network have equally good generalizing ability, such points have a huge generalization error. The paper studies the properties of such points and their location on the surface of the loss function of modern neural networks.

Training neural networks with verifiable robustness guarantees is challenging. Several existing successful approaches utilize relatively tight linear relaxation based bounds of neural network outputs, but they can slow down training by a factor of hundreds and over-regularize the network. Meanwhile, interval bound propagation (IBP) based training is efficient and significantly outperform linear relaxation based methods on some tasks, yet it suffers from stability issues since the bounds are much looser. In this paper, we first interpret IBP training as training an augmented network which computes non-linear bounds, thus explaining its good performance. We then propose a new certified adversarial training method, CROWN-IBP, by combining the fast IBP bounds in the forward pass and a tight linear relaxation based bound, CROWN, in the backward pass. The proposed method is computationally efficient and consistently outperforms IBP baselines on training verifiably robust neural networks. We conduct large scale experiments using 53 models on MNIST, Fashion-MNIST and CIFAR datasets. On MNIST with $\epsilon=0.3$ and $\epsilon=0.4$ ($\ell_\infty$ norm distortion) we achieve 7.46\% and 12.96\% verified error on test set, respectively, outperforming previous certified defense methods.

The unsupervised training of GANs and VAEs has enabled them to generate realistic images mimicking real-world distributions and perform image-based unsupervised clustering or semi-supervised classification. Combining the power of these two generative models, we introduce Multi-Adversarial Variational autoEncoder Networks (MAVENs), a novel network architecture that incorporates an ensemble of discriminators in a VAE-GAN network, with simultaneous adversarial learning and variational inference. We apply MAVENs to the generation of synthetic images and propose a new distribution measure to quantify the quality of the generated images. Our experimental results using datasets from the computer vision and medical imaging domains---Street View House Numbers, CIFAR-10, and Chest X-Ray datasets---demonstrate competitive performance against state-of-the-art semi-supervised models both in image generation and classification tasks.

This paper presents the experiments accomplished as a part of our participation in the MEDIQA challenge, an (Abacha et al., 2019) shared task. We participated in all the three tasks defined in this particular shared task. The tasks are viz. i. Natural Language Inference (NLI) ii. Recognizing Question Entailment(RQE) and their application in medical Question Answering (QA). We submitted runs using multiple deep learning based systems (runs) for each of these three tasks. We submitted five system results in each of the NLI and RQE tasks, and four system results for the QA task. The systems yield encouraging results in all three tasks. The highest performance obtained in NLI, RQE and QA tasks are 81.8%, 53.2%, and 71.7%, respectively.