AI has the power to search through millions of genetic variants at high speed and identify likely ... [ ] causes of rare diseases, while also comparing what they find with the existing medical literature. This is greater than the population of the United States, yet the ominous figures don't end there. According to the Global Genes organization, eight out of ten rare diseases are caused by a faulty gene, yet it takes an average of 4.8 years to arrive at an accurate diagnosis. This is part of the reason why 30% of children with a rare disease won't live to see their fifth birthday. Neither is this situation helped by the fact that 95% of rare diseases lack an FDA-approved treatment.

Amazon is rolling out new features that make it easier for customers to search for specific items in large databases of images and videos. The tech giant said Monday it will make "Custom Labels" available next month, allowing Rekognition users without any machine learning experience to deploy the software more nimbly. Custom Labels allow customers to train Rekognition to detect specific objects and scenes related to their use case. For example, a retailer who wants to identify all of the black boots in her online inventory could train the software with images of shoes that meet that description, so that it can detect them in a larger library of images. That type of custom sorting previously required machine learning expertise to build a model from scratch, an Amazon spokesperson said.

I'm sure you've seen the headlines: The artificial intelligence (AI) market is due to balloon in the coming years, and AI will change the world and solve problems we've never had answers to. While I believe that AI can certainly be useful and has some real-world applications already, I think a reality check of what AI does (and doesn't do) well is in order. The truth of AI is that production deployment lags behind claims from today's headlines. According to research conducted by MIT, for example, AI "Pioneers," which are defined as "organizations that both understand and have adopted AI," make up just 20% of all organizations. What is creating the disparity between theory and production?

DUBAI: We live in an age in which talking to a machine is no longer a futuristic concept. Many activities and fields are being shaped by innovations in artificial intelligence (AI) and technology, a phenomenon that is fundamentally altering the nature of human interaction, consumption of information and, more controversially, the nature of employment. Adoption of innovations such as voice recognition, virtual agents and biometrics could be the only way for humans to survive in the coming age of automation. Amid the surge in human-machine cooperation, many people nevertheless are eager to find out what the future holds. Will robots outsmart humans and steal their jobs?

The update shares an announcement from the United States Patent and Trademark Office ("USPTO") and provides important details regarding a second Federal Register Notice on AI and innovation building upon the earlier request for comments focused on the impact AI poses for patent law and policy. The Ocean Tomo AI Report Update contains important details regarding the window of opportunity to respond to the notice as well as sample questions included in the notice. You can request a copy of the study update here. Ocean Tomo Industry Analyst Reports provide a comprehensive look at current industry trends and deal activity in several technology areas. As a financial advisor with a focus on technology and intellectual property (IP), Ocean Tomo has gained unique insights related to the intellectual property driving the development of a variety of technology areas.

--Deep Neural Network (DNN) workloads are quickly moving from datacenters onto edge devices, for latency, privacy, or energy reasons. While datacenter networks can be protected using conventional cybersecurity measures, edge neural networks bring a host of new security challenges. Unlike classic IoT applications, edge neural networks are typically very compute and memory intensive, their execution is data-independent, and they are robust to noise and faults. Neural network models may be very expensive to develop, and can potentially reveal information about the private data they were trained on, requiring special care in distribution. The hidden states and outputs of the network can also be used in reconstructing user inputs, potentially violating users' privacy. Furthermore, neural networks are vulnerable to adversarial attacks, which may cause misclassifications and violate the integrity of the output. These properties add challenges when securing edge-deployed DNNs, requiring new considerations, threat models, priorities, and approaches in securely and privately deploying DNNs to the edge. In this work, we cover the landscape of attacks on, and defenses, of neural networks deployed in edge devices and provide a taxonomy of attacks and defenses targeting edge DNNs. Since the rise of deep learning in the last decade, many different libraries and frameworks for running and training deep neural networks (DNN) have been published and open-sourced. In that time, the landscape of software tools for training neural networks has moved from difficult-to-install libraries [1], and support for static graphs only [2], to industry-ready, easy-to-deploy frameworks [3], high-development efficiency [4], and support for dynamic graphs and justin-time compilation [5].

In environments with continuous state and action spaces, state-of-the-art actor-critic reinforcement learning algorithms can solve very complex problems, yet can also fail in environments that seem trivial, but the reason for such failures is still poorly understood. In this paper, we contribute a formal explanation of these failures in the particular case of sparse reward and deterministic environments. First, using a very elementary control problem, we illustrate that the learning process can get stuck into a fixed point corresponding to a poor solution. Then, generalizing from the studied example, we provide a detailed analysis of the underlying mechanisms which results in a new understanding of one of the convergence regimes of these algorithms. The resulting perspective casts a new light on already existing solutions to the issues we have highlighted, and suggests other potential approaches.

Pre-trained word embeddings encode general word semantics and lexical regularities of natural language, and have proven useful across many NLP tasks, including word sense disambiguation, machine translation, and sentiment analysis, to name a few. In supervised tasks such as multiclass text classification (the focus of this article) it seems appealing to enhance word representations with ad-hoc embeddings that encode task-specific information. We propose (supervised) word-class embeddings (WCEs), and show that, when concatenated to (unsupervised) pre-trained word embeddings, they substantially facilitate the training of deep-learning models in multiclass classification by topic. We show empirical evidence that WCEs yield a consistent improvement in multiclass classification accuracy, using four popular neural architectures and six widely used and publicly available datasets for multiclass text classification. Our code that implements WCEs is publicly available at https://github.com/AlexMoreo/word-class-embeddings

Recent advances in adversarial attacks uncover the intrinsic vulnerability of modern deep neural networks. Since then, extensive efforts have been devoted to enhancing the robustness of deep networks via specialized learning algorithms and loss functions. In this work, we take an architectural perspective and investigate the patterns of network architectures that are resilient to adversarial attacks. To obtain the large number of networks needed for this study, we adopt one-shot neural architecture search, training a large network for once and then finetuning the sub-networks sampled therefrom. The sampled architectures together with the accuracies they achieve provide a rich basis for our study. Our "robust architecture Odyssey" reveals several valuable observations: 1) densely connected patterns result in improved robustness; 2) under computational budget, adding convolution operations to direct connection edge is effective; 3) flow of solution procedure (FSP) matrix is a good indicator of network robustness. Based on these observations, we discover a family of robust architectures (RobNets). On various datasets, including CIFAR, SVHN, and Tiny-ImageNet, RobNets exhibit superior robustness performance to other widely used architectures. Notably, RobNets substantially improve the robust accuracy (~5% absolute gains) under both white-box and black-box attacks, even with fewer parameter numbers.

Significant advances have been made in artificial systems by using biological systems as a guide. However, there is often little interaction between computational models for emergent communication and biological models of the emergence of language. Many researchers in language origins and emergent communication take compositionality as their primary target for explaining how simple communication systems can become more like natural language. However, there is reason to think that compositionality is the wrong target on the biological side, and so too the wrong target on the machine-learning side. As such, the purpose of this paper is to explore this claim. This has theoretical implications for language origins research more generally, but the focus here will be the implications for research on emergent communication in computer science and machine learning---specifically regarding the types of programmes that might be expected to work and those which will not. I further suggest an alternative approach for future research which focuses on reflexivity, rather than compositionality, as a target for explaining how simple communication systems may become more like natural language. I end by providing some reference to the language origins literature that may be of some use to researchers in machine learning.