We study online convex optimization in the random order model, recently proposed by \citet{garber2020online}, where the loss functions may be chosen by an adversary, but are then presented to the online algorithm in a uniformly random order. Focusing on the scenario where the cumulative loss function is (strongly) convex, yet individual loss functions are smooth but might be non-convex, we give algorithms that achieve the optimal bounds and significantly outperform the results of \citet{garber2020online}, completely removing the dimension dependence and improving their scaling with respect to the strong convexity parameter. Our analysis relies on novel connections between algorithmic stability and generalization for sampling without-replacement analogous to those studied in the with-replacement i.i.d.~setting, as well as on a refined average stability analysis of stochastic gradient descent.

This post is co-authored by Lezgin Bakircioglu, Innovation and Security Manager at Daniel Wellington. In their own words, "Daniel Wellington (DW) is a Swedish fashion brand founded in 2011. Since its inception, it has sold over 11 million watches and established itself as one of the fastest-growing and most coveted brands in the industry." In this post, we share how DW saved 99% on translation costs with Amazon Translate and other AWS services. At DW, having the ability to respond to customers in their local language is critical to the customer journey.

The application of AI is very interesting and fantasizing in the domain of healthcare, defense, and security. However, we do not see a lot of commercial applications reaching end-users. The papers are promising but very difficult to implement on the scale, also because of the regulations. However, in the Retail domain, the stakes are not that high as healthcare, hence, can be experimented with. Retail encompasses shopping malls to online purchases and from transportation to stacking of items in warehouses.

Big-box retailer Walmart is using artificial intelligence (AI) to aid customers and personal shoppers and better handle still-surging online demand for groceries amidst the COVID-19 pandemic. In a blog post Thursday, Srini Venkatesan, Walmart's global tech executive vice president, noted that as Americans increasingly turned to the internet to shop for essentials, stores like Walmart were presented with a "unique challenge." The alternate shopping method combined with the volume of in-store shoppers – especially in the months of March and April – resulted in popular items quickly selling out. Last July, Walmart corporate affairs said the company had hired more than 400,000 new associates to mitigate the sudden "customer rush on essentials as lockdowns spread across the U.S." "Walmart's solution was to use artificial intelligence to help both customers and Personal Shoppers choose the best substitute for an out-of-stock item," said Venkatesan. An illustrated video included in the blog post shows a Walmart personal shopper who needs to make a substitution for an online order.

There is an escalating need for methods to identify latent patterns in text data from many domains. We introduce a new method to identify topics in a corpus and represent documents as topic sequences. Discourse Atom Topic Modeling draws on advances in theoretical machine learning to integrate topic modeling and word embedding, capitalizing on the distinct capabilities of each. We first identify a set of vectors ("discourse atoms") that provide a sparse representation of an embedding space. Atom vectors can be interpreted as latent topics: Through a generative model, atoms map onto distributions over words; one can also infer the topic that generated a sequence of words. We illustrate our method with a prominent example of underutilized text: the U.S. National Violent Death Reporting System (NVDRS). The NVDRS summarizes violent death incidents with structured variables and unstructured narratives. We identify 225 latent topics in the narratives (e.g., preparation for death and physical aggression); many of these topics are not captured by existing structured variables. Motivated by known patterns in suicide and homicide by gender, and recent research on gender biases in semantic space, we identify the gender bias of our topics (e.g., a topic about pain medication is feminine). We then compare the gender bias of topics to their prevalence in narratives of female versus male victims. Results provide a detailed quantitative picture of reporting about lethal violence and its gendered nature. Our method offers a flexible and broadly applicable approach to model topics in text data.

At many firms, the marketing function is rapidly embracing artificial intelligence. But in order to fully realize the technology's enormous potential, chief marketing officers must understand the various types of applications--and how they might evolve. Classifying AI by its intelligence level (whether it is simple task automation or uses advanced machine learning) and structure (whether it is a stand-alone application or is integrated into larger platforms) can help firms plan which technologies to pursue and when. Companies should take a stepped approach, starting with rule-based, stand-alone applications that help employees make better decisions, and over time deploying more-sophisticated and integrated AI systems in customer-facing situations. Of all a company's functions, marketing has perhaps the most to gain from artificial intelligence.

AWS customers are increasingly using computer vision (CV) models for improved efficiency and an enhanced user experience. For example, a live broadcast of sports can be processed in real time to detect specific events automatically and provide additional insights to viewers at low latency. Inventory inspection at large warehouses capture and process millions of images across their network to identify misplaced inventory. CV models can be built with multiple deep learning frameworks like TensorFlow, PyTorch, and Apache MXNet. These models typically have a large input payload of images or videos of varying size.

It's no surprise that the coronavirus pandemic has changed the way we shop, especially when it comes to groceries. Grocery delivery apps experienced a record number of downloads in March 2020, and by the following month, Walmart Grocery (which is now integrated into the Walmart app) surpassed Amazon as the No. 1 shopping app on both Google Play and the App Store. But even as pandemic restrictions have eased, consumers are still using ordering groceries for delivery or pickup more frequently than they were pre-pandemic. As Walmart's grocery delivery services have continued to boom, posing competition to companies like Amazon and Instacart, the tech that Walmart uses has expanded too. Today, Walmart shared information about how it's training its AI to make smarter substitutions in online grocery orders.

Last AWS re:Invent, we announced the general availability of Amazon EMR on Amazon Elastic Kubernetes Service (Amazon EKS), a new deployment option for Amazon EMR that allows customers to automate the provisioning and management of Apache Spark on Amazon EKS. With Amazon EMR on EKS, customers can deploy EMR applications on the same Amazon EKS cluster as other types of applications, which allows them to share resources and standardize on a single solution for operating and managing all their applications. Customers running Apache Spark on Kubernetes can migrate to EMR on EKS and take advantage of the performance-optimized runtime, integration with Amazon EMR Studio for interactive jobs, integration with Apache Airflow and AWS Step Functions for running pipelines, and Spark UI for debugging. When customers submit jobs, EMR automatically packages the application into a container with the big data framework and provides prebuilt connectors for integrating with other AWS services. EMR then deploys the application on the EKS cluster and manages running the jobs, logging, and monitoring.

Retail, one of the largest industries in the world, totaled an estimated $5.73 trillion in sales in 2017 in the US according to Plunkett Research. This is an increase compared to the $5.52 trillion in sales in 2016 according to the U.S Census Bureau. With the help of artificial intelligence, consumers are experiencing a whole new way of shopping both online and in stores. AI in retail is being applied in innovative ways - from before a sale to after products or services is purchased. Businesses of tomorrow must keep up with customer expectations and implementing AI can help.