Breakthroughs in the application of complex calculations to large volumes of data have enabled machine-learning methodologies to revolutionize business processes in nearly every industry. Some of the more recognized examples of machine-learning applications include personalized Netflix recommendations and related product modules from online retailers such as Amazon and Nordstrom. However, there are less sexy yet equally impactful machine-learning examples, which include revenue management solutions used in hotels that incorporate these methodologies into an algorithmic engine to help produce pricing and inventory recommendations. Unlocking the potential of machine learning for the office of finance remains a hot topic for financial planning and analysis (FP&A) leaders, industry analysts, and technology vendors alike. Even more specifically, continuous chatter surrounds the ways that machine learning can improve future FP&A processes and how finance leaders can prepare for deploying advanced analytics within their organizations.

With ML (machine learning), algorithms rewrite themselves as the machine "learns" more and more about patient care. I've written about a recent case where IBM's Watson computer "read" a Japanese leukemia patient's medical records, genetic data, and 20 million journal articles on leukemia (all in around 10 minutes) and concluded that teams of doctors had misdiagnosed her illness and treated her with the wrong medications. Watson effectively, continually reprogrammed itself to analyze the patient's illness in ways no human had done.

To what extent can your doctor's functions be automated -- replaced or enhanced by intelligent machines? How might such automation improve care and reduce costs? These questions are central to understanding Clover Health -- a California-based company providing Medicare Advantage insurance plans in seven states: New Jersey, Pennsylvania, Tennessee, Georgia, Arizona, South Carolina and Texas. A while back, I hosted a dinner in New York for a dozen-plus health care innovators -- entrepreneurs, medical school professors, futurists, etc. Someone in the room asked, "How much of today's physician services can be reduced to algorithms?" An algorithm is a set of instructions (like a computer program) leading to unambiguous results.

In my experience as a C-level executive and long-time AI professional, I've learned that people who want to utilize artificial Intelligence find getting started to be the most difficult part. Even the more confident practitioners could easily become intimidated by the array and complexity of tools to navigate. But this problem is now a thing of the past. With IBM Watson Studio, you and your project can now hit the ground running. IBM Watson Studio's integrated environment makes AI significantly easier, by allowing users to quickly and easily build visually appealing projects and models. I don't have the luxury to get bogged down in inefficient and slow processes.

For most businesses, machine learning seems close to rocket science, appearing expensive and talent demanding. And, if you're aiming at building another Netflix recommendation system, it really is. But the trend of making everything-as-a-service has affected this sophisticated sphere, too. You can jump-start an ML initiative without much investment, which would be the right move if you are new to data science and just want to grab the low hanging fruit. One of ML's most inspiring stories is the one about a Japanese farmer who decided to sort cucumbers automatically to help his parents with this painstaking operation. Unlike the stories that abound about large enterprises, the guy had neither expertise in machine learning, nor a big budget. But he did manage to get familiar with TensorFlow and employed deep learning to recognize different classes of cucumbers. By using machine learning cloud services, you can start building your first working models, yielding valuable insights from predictions with a relatively small team. We've already discussed machine learning strategy. Now let's have a look at the best machine learning platforms on the market and consider some of the infrastructural decisions to be made.

After logging into Watson Studio, select New Modeler Flow. Enter a name, keep the default settings, and then click Create. Next expand the Import menu, drag the Data Asset node onto the stream canvas and select Titanic training data file (train.csv) in the node settings to load data into the project. Right-click the node and select Preview to see your detailed dataset. To build a modeler stream look under Record Operations.

Memorization of data in deep neural networks has become a subject of significant research interest. In this paper, we link memorization of images in deep convolutional autoencoders to downsampling through strided convolution. To analyze this mechanism in a simpler setting, we train linear convolutional autoencoders and show that linear combinations of training data are stored as eigenvectors in the linear operator corresponding to the network when downsampling is used. On the other hand, networks without downsampling do not memorize training data. We provide further evidence that the same effect happens in nonlinear networks. Moreover, downsampling in nonlinear networks causes the model to not only memorize linear combinations of images, but individual training images. Since convolutional autoencoder components are building blocks of deep convolutional networks, we envision that our findings will shed light on the important phenomenon of memorization in over-parameterized deep networks.

In a digital business environment, providing a quality customer experience -- on multiple digital fronts -- is not only a crucial aspect in modern business strategies, but it's also becominga key responsibility of the CIO. AI and machine learning tools have a significant role to play. According to Gartner, customer experience (CX) represents the majority of AI businessvalue through 2020. AI-driven customer experience projects are still nascent, however. AGartner survey found that 50%of customer experience professionals are using digital analytics or big data in their CRM/CX projects, but only 26% are using AI or machine learning.

IBM's Watson supercomputer has beat Jeopardy champions, reconstituted recipes, and even helped create highlight reels for the World Cup. Now it's taking on a new tech challenge; changing how the construction industry operates. A new partnership between IBM and Fluor, a global engineering and construction company, will put the supercomputer's computational skills to work on making building more efficient. The new Watson-based system, in development since 2015 and now in use on select projects, will be able to analyze a job site "like a doctor diagnoses a patient," according to Leslie Lindgren, Fluor's vice president of Information Management. That degree of risk analysis, predictive logistics, and comprehension is no small challenge given the complexity of today's construction megaprojects.

We propose a new notion of'non-linearity' of a network layer with respect to an input batch that is based on its proximity to a linear system, which is reflected in the nonnegative rank of the activation matrix. Considering batches of similar samples, we find that high non-linearity in deep layers is indicative of memorization. Furthermore, by applying our approach layer-by-layer, we find that the mechanism for memorization consists of distinct phases. We perform experiments on fully-connected and convolutional neural networks trained on several image and audio datasets. Our results demonstrate that as an indicator for memorization, our technique can be used to perform early stopping. A fundamental challenge in machine learning is balancing the bias-variance tradeoff, where overly simple learning models underfit the data (suboptimal performance on the training data) and overly complex models are expected to overfit or memorize the data (perfect training set performance, but suboptimal test set performance). The latter direction of this tradeoff has come into question with the observation that deep neural networks do not memorize their training data despite having sufficient capacity to do so (Zhang et al., 2016), the explanation of which is a matter of much interest. Due to their convenient gradient properties and excellent performance in practice, rectified-linear units (ReLU) have been widely adopted and are now ubiquitous in the field of deep learning.