When Microsoft CEO Satya Nadella visited New Zealand three years ago, it was an event on the tech calendar. Friday's Translator announcement made his latest visit a cultural milestone. More than 1000 industry leaders and media filled the room at Auckland's Eden Park where Satya announced the addition of te reo Māori to Microsoft Translator, alongside the 60 languages already supported by the free application. The move will enable anyone around the world to translate text to and from te reo Māori. At a time when just 3 per cent of Kiwis speak te reo Māori, yet Prime Minister Jacinda Ardern is calling for 1 million additional speakers by 2040, it's crucial that people have the tools to engage with each other in te reo Māori in their everyday lives.

The United States government says it will limit the export of certain types of artificial intelligence software that's used to analyze images from satellites in order to keep it out of the hands of foreign rivals such as China. Reuters said the ban, which goes into effect Monday, relates to a 2018 law known as the Export Control Reform Act that requires the government look into how it can restrict the export of new technologies it deems "essential to the national security" of the U.S. The scope of the ban is rather narrow, at least for now. It applies specifically to software that uses neural networks, a component of machine learning, to discover "points of interest" in geospatial images created by satellites. For example, software that can identify houses or vehicles. Furthermore, the ban only applies to software that has a graphical user interface.

The "LegalTech Artificial Intelligence Market" is evolving at an exciting pace driven by changing dynamics and risk ecosystem, an analysis of which forms the crux of the report. The study on the global LegalTech Artificial Intelligence Market takes a closer look at several regional trends and the emerging regulatory landscape to assess its prospects. The critical evaluation of the various growth factors and opportunities in the global LegalTech Artificial Intelligence Market offered in the analyses helps in assessing the lucrativeness of its key segments. Summary of Market: The global LegalTech Artificial Intelligence market is valued at xx million US$ in 2019 is expected to reach xx million US$ by the end of 2025, growing at a CAGR of xx% during 2019-2025. Legal technology, also known asLegal Tech, refers to the use oftechnologyandsoftwareto providelegal services.

Three police officers were assaulted and six juveniles were arrested after multiple fights broke out at a Mesa skate rink on Saturday night, authorities said. Weekend wrap-up: Here are the biggest Arizona stories from Jan. 3-5 Blowback: Iran abandons nuclear limits after US killing Luxury garage storage company bringing 2 new locations to the Valley Weinstein's reckoning: Trial looms 2 years after #MeToo wave Valley eye surgeon faces multiple charges for alleged billing scheme Weekend wrap-up: Here are the biggest Arizona stories from Jan. 3-5 Counting whales from space pitched as key to saving them Iraq's Parliament calls for expulsion of U.S. troops Tips on how to create, manage your budget in the new year Iraq's Parliament calls for expulsion of U.S. troops Iraq's Parliament calls for expulsion of U.S. troops Arizona ex-fire chief pleads guilty to theft charges A former fire chief accused of embezzling $40,000 from his Arizona district pleaded guilty to felony charges of theft. Valley doctor says soot from candles can be harmful to your health Candles smell good but they're not all that great for your health for one specific reason, according to one Valley doctor. Phoenix lab uses artificial intelligence to slow, manage Alzheimer's disease Arizona is projected to have one of the fastest growing rates of Alzheimer's disease in the country over the next few years, and a clinical lab testing company in the Valley is trying to reverse that. '1917,' 'Once Upon a Time ...in Hollywood' win Golden Globes See winners from the 2020 Golden Globes, hosted by Ricky Gervais.

For the past few years, we've shared predictions each December on what we believe will be the top ten technology policy issues for the year ahead. As this year draws to a close, we are looking out a bit further. It gives us all an opportunity to reflect upon the past ten years and consider what the 2020s may bring. As we concluded in our book, Tools and Weapons: The Promise and the Peril of the Digital Age, "Technology innovation is not going to slow down. The work to manage it needs to speed up." Digital technology has gone longer with less regulation than virtually any major technology before it. This dynamic is no longer sustainable, and the tech sector will need to step up and exercise more responsibility while governments catch up by modernizing tech policies. In short, the 2020s will bring sweeping regulatory changes to the world of technology. Tech is at a crossroads, and to consider why, it helps to start with the changes in technology itself. The 2010s saw four trends intersect, collectively transforming how we work, live and learn. Continuing advances in computational power made more ambitious technical scenarios possible both for devices and servers, while cloud computing made these advances more accessible to the world. Like the invention of the personal computer itself, cloud computing was as important economically as it was technically. The cloud allows organizations of any size to tap into massive computing and storage capacity on demand, paying for the computing they need without the outlay of capital expenses. More powerful computers and cloud economics combined to create the third trend, the explosion of digital data.

The vulnerability of deep neural networks to adversarial attacks has been widely demonstrated (e.g., adversarial example attacks). Traditional attacks perform unstructured pixel-wise perturbation to fool the classifier. An alternative approach is to have perturbations in the latent space. However, such perturbations are hard to control due to the lack of interpretability and disentanglement. In this paper, we propose a more practical adversarial attack by designing structured perturbation with semantic meanings. Our proposed technique manipulates the semantic attributes of images via the disentangled latent codes. The intuition behind our technique is that images in similar domains have some commonly shared but theme-independent semantic attributes, e.g. thickness of lines in handwritten digits, that can be bidirectionally mapped to disentangled latent codes. We generate adversarial perturbation by manipulating a single or a combination of these latent codes and propose two unsupervised semantic manipulation approaches: vector-based disentangled representation and feature map-based disentangled representation, in terms of the complexity of the latent codes and smoothness of the reconstructed images. We conduct extensive experimental evaluations on real-world image data to demonstrate the power of our attacks for black-box classifiers. We further demonstrate the existence of a universal, image-agnostic semantic adversarial example.

Language Models Are An Effective Patient Representation Learning Technique For Electronic Health Record Data Ethan Steinberg, Ken Jung, Jason A. Fries, Conor K. Corbin, Stephen R. Pfohl, Nigam H. Shah January 16, 2020 Abstract Widespread adoption of electronic health records (EHRs) has fueled development of clinical outcome models using machine learning. However, patient EHR data are complex, and how to optimally represent them is an open question. This complexity, along with often small training set sizes available to train these clinical outcome models, are two core challenges for training high quality models. In this paper, we demonstrate that learning generic representations from the data of all the patients in the EHR enables better performing prediction models for clinical outcomes, allowing for these challenges to be overcome. We adapt common representation learning techniques used in other domains and find that representations inspired by language models enable a 3.5% mean improvement in AUROC on five clinical outcomes compared to standard baselines, with the average improvement rising to 19% when only a small number of patients are available for training a prediction model for a given clinical outcome. 1 Introduction The widespread adoption of electronic health records (EHRs) has created opportunities for using machine learning to reduce healthcare costs and improve quality of care. EHR data have been used to learn prediction models for clinical outcomes such as mortality [1], sepsis [2], 30-day readmission [3] and others [4, 5]. However, the complexity of patient data poses many obstacles to its effective use. Patient records in EHRs are variable length, high dimensional and sparse, with complex temporal and hierarchical structure. They are comprised of irregularly spaced visits spread across years, with each visit consisting of a subset of thousands of possible diagnosis, procedure, and medication codes as well as lab values and unstructured data such as text or images. In contrast, most off-the-shelf machine learning algorithms expect a fixed length vector of features as input. Manually defining a transformation of patient records into such a representation beyond simple binned counts is time consuming and outcome-dependent, leaving much of the temporal and hierarchical structure of EHRs underutilized when building machine learning models. The challenge of representing EHR data can be addressed by using neural networks to automatically learn how to featurize patient data while learning a model for a given clinical outcome (e.g., mortality or 30 day readmissions) [4].

Federated learning is an emerging research paradigm to train models on private data distributed over multiple devices. A key challenge involves keeping private all the data on each device and training a global model only by communicating parameters and updates. Overcoming this problem relies on the global model being sufficiently compact so that the parameters can be efficiently sent over communication channels such as wireless internet. Given the recent trend towards building deeper and larger neural networks, deploying such models in federated settings on real-world tasks is becoming increasingly difficult. To this end, we propose to augment federated learning with local representation learning on each device to learn useful and compact features from raw data. As a result, the global model can be smaller since it only operates on higher-level local representations. We show that our proposed method achieves superior or competitive results when compared to traditional federated approaches on a suite of publicly available real-world datasets spanning image recognition (MNIST, CIFAR) and multimodal learning (VQA). Our choice of local representation learning also reduces the number of parameters and updates that need to be communicated to and from the global model, thereby reducing the bottleneck in terms of communication cost. Finally, we show that our local models provide flexibility in dealing with online heterogeneous data and can be easily modified to learn fair representations that obfuscate protected attributes such as race, age, and gender, a feature crucial to preserving the privacy of on-device data.

With the success of deep neural networks, Neural Architecture Search (NAS) as a way of automatic model design has attracted wide attention. As training every child model from scratch is very time-consuming, recent works leverage weight-sharing to speed up the model evaluation procedure. These approaches greatly reduce computation by maintaining a single copy of weights on the super-net and share the weights among every child model. However, weight-sharing has no theoretical guarantee and its impact has not been well studied before. In this paper, we conduct comprehensive experiments to reveal the impact of weight-sharing: (1) The best-performing models from different runs or even from consecutive epochs within the same run have significant variance; (2) Even with high variance, we can extract valuable information from training the super-net with shared weights; (3) The interference between child models is a main factor that induces high variance; (4) Properly reducing the degree of weight sharing could effectively reduce variance and improve performance.

Its impact is drastic and real: Youtube's AIdriven recommendation system would present sports videos for days if one happens to watch a live baseball game on the platform [1]; email writing becomes much faster with machine learning (ML) based auto-completion [2]; many businesses have adopted natural language processing based chatbots as part of their customer services [3]. AI has also greatly advanced human capabilities in complex decision-making processes ranging from determining how to allocate security resources to protect airports [4] to games such as poker [5] and Go [6]. All such tangible and stunning progress suggests that an "AI summer" is happening. As some put it, "AI is the new electricity" [7]. Meanwhile, in the past decade, an emerging theme in the AI research community is the so-called "AI for social good" (AI4SG): researchers aim at developing AI methods and tools to address problems at the societal level and improve the wellbeing of the society.