Since its launch in 1987, the annual Conference on Neural Information Processing Systems (NIPS) has brought together researchers working on neural networks and related fields, but it later diversified to become one of the largest conferences in machine learning. In recent years, the trend towards deep learning has brought the conference closer to its roots. The 2016 program spanned six days (Dec 5 to 10) and included tutorials, oral and poster presentations, workshops, and invited talks on a broad range of research topics. Following their previous Insights post on ICML 2016, Two Sigma researchers Vinod Valsalam and Firdaus Janoos discuss below the notable advances in deep learning, optimization algorithms, Bayesian techniques, and time-series analysis presented at NIPS 2016. With 550 accepted papers and 50 workshops, the number of attendees more than doubled in the past two years (from more than 2500 in 2014 to over 5000 in 2016), demonstrating rapidly growing interest in machine learning and artificial intelligence. That included strong industry participation (Two Sigma was among the more than 60 sponsors), both for recruiting talent as well as for presenting advances in the field. Several interesting invited talks were given by researchers who are established in both academia and industry.

Will artificial intelligence (AI) replace radiologists? During a session on AI and imaging yesterday at the Big Data in Biomedicine conference, panelists preempted this question (which keeps some radiologists up at night) by clarifying how, at least for now, AI isn't a replacement for doctors, but a tool to help them. "The human-machine system always performs better than either alone," said Curt Langlotz, MD, PhD, a professor of radiology and biomedical informatics at Stanford. And while AI is achieving human-level performance, it's not necessarily superseding it -- yet. All panelists spoke about AI's capacity to increase efficiency.

Data is the fuel for AI, and Google owns some of the largest data sets in the world. The company operates seven services with over a billion monthly active users: Android, Chrome, YouTube, Gmail, Google Maps, Google Search, and Google Play. In addition, Google Translate and Google Photos are used by over 500 million people each. By operating such a diverse range of services, Google collects data of various types: text, images, video, maps, and webpages--helping the company master not just one kind of AI, but AI across various use cases. Just as important as the data are the apps, which Google also owns.

Once again, artificial intelligence triumphed over man. In the second match of a three-game series on Thursday, Google's DeepMind AlphaGo program beat the 19-year-old Chinese prodigy Ke Jie in the strategy board game Go. AlphaGo won the first game earlier in the week; the final game is scheduled for Saturday. The daily Bits newsletter will keep you updated on the latest from Silicon Valley and the technology industry, plus exclusive analysis from our reporters and editors. Please verify you're not a robot by clicking the box. You must select a newsletter to subscribe to.

According to a new market research report by MarketsandMarkets, the market is expected to grow from $667.1 million in 2016 to $7,988.8 million by 2022, at a CAGR of 52.68 percent during the forecast period. The growing usage of big data in the healthcare industry, ability of AI to improve patient outcomes, imbalance between health workforce and patients, reducing the healthcare costs, growing importance on precision medicine, cross-industry partnerships, and significant increase in venture capital investments are expected to drive the AI in healthcare market. Software to hold the largest share of the AI in healthcare market The AI software is used to assist the medical system in relevant insights, medical imaging and diagnostics, drug discovery, in-patient care and hospital management, virtual assistance, precision medicine, lifestyle management and monitoring, patient data and risk analysis, and research. The growing usage of smart devices, and the presence of major AI software providers such as IBM Corporation (US), Google Inc. (US) and Microsoft Corporation (US), Enlitic, Inc. (US), Next IT Corp (US) are driving the growth of the AI in healthcare market for the software offering. Deep learning technology expected to grow at the highest rate between 2017 and 2022 The deep learning technology which includes image recognition, signal recognition, and data mining-is expected to witness the highest CAGR during the forecast period.

Through deep learning, algorithms and data help machines improve their accuracy and knowledge over time, which means we don't have to do the grunt work normally required to be strategic because the program is doing it for us. I can't even begin to estimate the number of hours I spent weekly, monthly, or even yearly sorting in Excel, navigating clunky databases, and Googling names, while trying to be more intentional with my outreach and travel--basically, spending time to save time. For decades, we advancement folk have been in pursuit of that perfect but elusive combination of skill and luck that will direct us toward the most promising donors at a given time. Machines can now do this for us (well, Gravyty can), just like Facebook can tell us who to tag in photos through face recognition--but identifying donors is way more helpful and lucrative than that.

Last November I wrote a column about IBM's newly announced AI software toolkit, PowerAI. PowerAI fulfills a special niche part of IBM's AI investment--aimed towards enterprises who want something in between Watson's turnkey solution ("easy button" on IBM Cloud) and a totally DIY on-prem infrastructure or public cloud. PowerAI runs on IBM's highest performing server in its OpenPOWER LC line (the Power S822LC for High Performance Computing), and utilizes deep learning frameworks and building block software to make it easier for enterprises to dive into AI and machine learning. Already these systems are landing in IT research groups and business units at companies. Recently, IBM announced a significant revamp of PowerAI, seeking to address some of the bigger challenges facing developers and data scientists--cutting down the time required for AI system training significantly, and simplifying the development experience.

While the geometry of classification regions and decision functions induced by traditional classifiers (such as linear and kernel SVM) is fairly well understood, these fundamental geometric properties are to a large extent unknown for state-of-the-art deep neural networks. Yet, to understand the recent success of deep neural networks and potentially address their weaknesses (such as their instability to perturbations [1]), an understanding of these geometric properties remains primordial. While many fundamental properties of deep networks have recently been studied, such as their optimization landscape in [2], [3], their generalization in [4], [5], and their expressivity in [6], [7], the geometric properties of the decision boundary and classification regions of deep networks has comparatively received little attention. The goal of this paper is to analyze these properties, and leverage them to improve the robustness of such classifiers to perturbations. In this paper, we specifically view classification regions as topological spaces, and decision boundaries as hypersurfaces and examine their geometric properties. We first study the classification regions induced by state-of-the-art deep networks, and provide empirical evidence suggesting that these classification regions are connected; that is, there exists a continuous path that remains in the region between any two points of the same label. Up to our knowledge, this represents the first instance where the connectivity of classification regions is empirically shown. Then, to study the complexity of the functions learned by the deep network, we analyze the curvature of their decision boundary.

We propose a novel technique to make neural network robust to adversarial examples using a generative adversarial network. We alternately train both classifier and generator networks. The generator network generates an adversarial perturbation that can easily fool the classifier network by using a gradient of each image. Simultaneously, the classifier network is trained to classify correctly both original and adversarial images generated by the generator. These procedures help the classifier network to become more robust to adversarial perturbations. Furthermore, our adversarial training framework efficiently reduces overfitting and outperforms other regularization methods such as Dropout. We applied our method to supervised learning for CIFAR datasets, and experimental results show that our method significantly lowers the generalization error of the network. To the best of our knowledge, this is the first method which uses GAN to improve supervised learning.

NVIDIA Volta Architecture – Volta is the world's most powerful GPU computing architecture, created to drive the next wave of advancement in AI and high performance computing. The first Volta-based processor is the Tesla V100 data center GPU, which brings extraordinary speed and scalability for AI inferencing and training, as well as for accelerating HPC and graphics workloads. New Volta-Based DGX Systems -- The company announced a new lineup of NVIDIA DGX AI supercomputers with unmatched computing performance. Using NVIDIA Tesla V100 data center GPUs based on the new Volta architecture and a fully optimized AI software package, the systems deliver groundbreaking AI computing power three times faster than the prior DGX generation, providing the performance of up to 800 CPUs in a single system. NVIDIA GPU Cloud Platform – NVIDIA GPU Cloud (NGC) is a cloud-based platform that will give developers convenient access -- via their PC, DGX system or the cloud -- to a comprehensive software suite for harnessing the transformative powers of AI.