Attacks against the control processor of a power-grid system, especially zero-day attacks, can be catastrophic. Earlier detection of the attacks can prevent further damage. However, detecting zero-day attacks can be challenging because they have no known code and have unknown behavior. In order to address the zero-day attack problem, we propose a data-driven defense by training a temporal deep learning model, using only normal data from legitimate processes that run daily in these power-grid systems, to model the normal behavior of the power-grid controller. Then, we can quickly find malicious codes running on the processor, by estimating deviations from the normal behavior with a statistical test. Experimental results on a real power-grid controller show that we can detect anomalous behavior with over 99.9% accuracy and nearly zero false positives.

In this work, we propose an ensemble of classifiers to distinguish between various degrees of abnormalities of the heart using Phonocardiogram (PCG) signals acquired using digital stethoscopes in a clinical setting, for the INTERSPEECH 2018 Computational Paralinguistics (ComParE) Heart Beats Sub-Challenge. Our primary classification framework constitutes a convolutional neural network with 1D-CNN time-convolution (tConv) layers, which uses features transferred from a model trained on the 2016 Physionet Heart Sound Database. We also employ a Representation Learning (RL) approach to generate features in an unsupervised manner using Deep Recurrent Autoencoders and use Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA) classifiers. Finally, we utilize an SVM classifier on a high-dimensional segment-level feature extracted using various functionals on short-term acoustic features, i.e., Low-Level Descriptors (LLD). An ensemble of the three different approaches provides a relative improvement of 11.13% compared to our best single subsystem in terms of the Unweighted Average Recall (UAR) performance metric on the evaluation dataset.

Neural networks designed for the task of classification have become a commodity in recent years. Many works target the development of better networks, which results in a complexification of their architectures with more layers, multiple sub-networks, or even the combination of multiple classifiers. In this paper, we show how to redesign a simple network to reach excellent performances, which are better than the results reproduced with CapsNet on several datasets, by replacing a layer with a Hit-or-Miss layer. This layer contains activated vectors, called capsules, that we train to hit or miss a central capsule by tailoring a specific centripetal loss function. We also show how our network, named HitNet, is capable of synthesizing a representative sample of the images of a given class by including a reconstruction network. This possibility allows to develop a data augmentation step combining information from the data space and the feature space, resulting in a hybrid data augmentation process. In addition, we introduce the possibility for HitNet, to adopt an alternative to the true target when needed by using the new concept of ghost capsules, which is used here to detect potentially mislabeled images in the training data.

We study the challenges of applying deep learning to gene expression data. We find experimentally that there exists non-linear signal in the data, however is it not discovered automatically given the noise and low numbers of samples used in most research. We discuss how gene interaction graphs (same pathway, protein-protein, co-expression, or research paper text association) can be used to impose a bias on a deep model similar to the spatial bias imposed by convolutions on an image. We explore the usage of Graph Convolutional Neural Networks coupled with dropout and gene embeddings to utilize the graph information. We find this approach provides an advantage for particular tasks in a low data regime but is very dependent on the quality of the graph used. We conclude that more work should be done in this direction. We design experiments that show why existing methods fail to capture signal that is present in the data when features are added which clearly isolates the problem that needs to be addressed.

Not a week goes by that I don't see dozens of headlines blaming "AI" or "deep learning" for yet another ethical conundrum that will doom human society. Whether it is predictive policing or facial recognition or autonomous weapons, it seems nearly every facet of society is facing an "AI revolution" that will destroy humankind. If one peels back the breathless hype and viral buzzwords, however, is it really AI that we are worried about or is it the shift towards a data-centric society with or without deep learning advances? To the general public, "big data" and "deep learning" are increasingly becoming synonymous, fueled by the never-ending hype machine of Silicon Valley and the very legitimate advances occurring in deep learning powered largely by the massive availability of large datasets and the unique abilities of those models to make sense of all that data. From a technical standpoint, however, these are two entirely distinct concepts.

It all starts with a language model. Let's assume we have the sequence [my, cat's, breath, smells, like, cat, ____] and we want to guess the final word. There are several ways to create a language model. The most straightforward is an n-gram model that counts occurrences to estimate frequencies. A bare-bones implementation requires only a dozen lines of Python code and can be surprisingly powerful.

A few days ago, Amazon announced the availability of a new set of automatic model tuning capabilities in the AWS SageMaker platform. Specifically, the new releases focuses on tuning and optimizing hyperparameters associated with SageMaker models. The release constitutes a powerful addition in order to streamline the adoption of SageMaker within the data science community. With the new hyperparameter tuning and model optimization capabilities, SageMaker joins a new group of platforms that are entering the market trying to solve this notorious challenge in deep learning applications. Building deep learning solutions in the real world is a process of constant experimentation and optimization.

Our brains are wired in a way that they can differentiate between objects, both living and non-living by simply looking at them. In fact, the recognition of objects and a situation through visualization is the fastest way to gather, as well as to relate information. This becomes a pretty big deal for computers where a vast amount of data has to be stuffed into it, before the computer can perform an operation on its own. Ironically, with each passing day, it is becoming essential for machines to identify objects through facial recognition, so that humans can take the next big step towards a more scientifically advanced social mechanism. So, what progress have we really made in that respect?

We are now in the middle of 2018 and Deep Learning research is advancing at exponential rates. At the beginning of this year, I made 10 predictions on what to expect for the year. Making predictions and comparing them in hindsight is one way to determine if ones expectations are overshooting reality. It turns out, it's undershooting reality in one aspect. I had not expected to see this much new research in "self-awareness". It's a consensus understanding that self-awareness in machines can lead to more autonomous machines and ultimately into machines with consciousness.

Artificial intelligence can do just about anything these days, like produce 3D renderings of an object from snapshots, defeat facial recognition systems, or track wildlife in the Serengeti. In a white paper titled "Dank Learning" (yes, really), Abel L. Peirson and E. Meltem Tolunay, the two lead scientists on the project, describe a neural network that ingests, gains an understanding of, and spits out internet in-jokes. The AI consists of a convolutional neural network (CNN) that takes images as inputs and translates them into mathematical representations called vector embeddings (an encoder), and a long short-term memory (LSTM) recurrent neural network (RNN) that creates captions (a decoder). Here's a tricky question: Can you tell which of the following memes were created by the neural net? The Stanford researchers fed the system more than 400,000 images with 2,600 unique image label-pairs from memegenerator.com -- specifically "advice animal"-style memes, pictures with humorously captioned specific characters (for example, a cat in a bathrobe) -- using a Python script.