While gender bias in large language models (LLMs) has been extensively studied in many domains, uses of LLMs in e-commerce remain largely unexamined and may reveal novel forms of algorithmic bias and harm. Our work investigates this space, developing data-driven taxonomic categories of gender bias in the context of product description generation, which we situate with respect to existing general purpose harms taxonomies. We illustrate how AI-generated product descriptions can uniquely surface gender biases in ways that require specialized detection and mitigation approaches. Further, we quantitatively analyze issues corresponding to our taxonomic categories in two models used for this task -- GPT-3.5 and an e-commerce-specific LLM -- demonstrating that these forms of bias commonly occur in practice. Our results illuminate unique, under-explored dimensions of gender bias, such as assumptions about clothing size, stereotypical bias in which features of a product are advertised, and differences in the use of persuasive language. These insights contribute to our understanding of three types of AI harms identified by current frameworks: exclusionary norms, stereotyping, and performance disparities, particularly for the context of e-commerce.

Abstract-- We present a virtual reality (VR) framework to automate the data collection process in cloth folding tasks. The framework uses skeleton representations to help the user define the folding plans for different classes of garments, allowing for replicating the folding on unseen items of the same class. We evaluate the framework in the context of automating garment folding tasks. A quantitative analysis is performed on three classes of garments, demonstrating that the framework reduces the need for intervention by the user. We also compare skeleton representations with RGB images in a classification task on a large dataset of clothing items, motivating the use of the proposed framework for other classes of garments.

Image-based virtual try-on aims to synthesize an image of a person wearing a given clothing item. To solve the task, the existing methods warp the clothing item to fit the person's body and generate the segmentation map of the person wearing the item before fusing the item with the person. However, when the warping and the segmentation generation stages operate individually without information exchange, the misalignment between the warped clothes and the segmentation map occurs, which leads to the artifacts in the final image. The information disconnection also causes excessive warping near the clothing regions occluded by the body parts, so-called pixel-squeezing artifacts. To settle the issues, we propose a novel try-on condition generator as a unified module of the two stages (i.e., warping and segmentation generation stages). A newly proposed feature fusion block in the condition generator implements the information exchange, and the condition generator does not create any misalignment or pixel-squeezing artifacts. We also introduce discriminator rejection that filters out the incorrect segmentation map predictions and assures the performance of virtual try-on frameworks. Experiments on a high-resolution dataset demonstrate that our model successfully handles the misalignment and occlusion, and significantly outperforms the baselines. Code is available at https://github.com/sangyun884/HR-VITON.

In three laboratory experiments, we examine the impact of personally relevant failures (PeRFs) on perceptions of a collaborative robot. PeR is determined by how much a specific issue applies to a particular person, i.e., it affects one's own goals and values. We hypothesized that PeRFs would reduce trust in the robot and the robot's Likeability and Willingness to Use (LWtU) more than failures that are not personal to participants. To achieve PeR in human-robot interaction, we utilized three different manipulation mechanisms: A) damage to property, B) financial loss, and C) first-person versus third-person failure scenarios. In total, 132 participants engaged with a robot in person during a collaborative task of laundry sorting. All three experiments took place in the same experimental environment, carefully designed to simulate a realistic laundry sorting scenario. Results indicate that the impact of PeRFs on perceptions of the robot varied across the studies. In experiments A and B, the encounters with PeRFs reduced trust significantly relative to a no failure session. But not entirely for LWtU. In experiment C, the PeR manipulation had no impact. The work highlights challenges and adjustments needed for studying robotic failures in laboratory settings. We show that PeR manipulations affect how users perceive a failing robot. The results bring about new questions regarding failure types and their perceived severity on users' perception of the robot. Putting PeR aside, we observed differences in the way users perceive interaction failures compared (experiment C) to how they perceive technical ones (A and B).

Imagine: social distancing restrictions are over. It's safe(r) to go out again! For once, after a long, grueling era of pandemic stress, you make plans to go out to a special public event. It hasn't happened for the longest time. Clearly, this is a cause for celebration and what else to mark the occasion than to dress yourself up a little?

Roblox wants to make its avatars look less blocky and more realistic, and it has announced a couple of visual updates meant to achieve that goal during its annual developers conference. One of those changes is layered clothing, which it's been working on since at least 2020. It allows any type of character model to be outfitted with layered clothing items. TechCrunch explains that the feature ensures clothing items will fit avatars and will drape around them naturally, whether they're human- or dinosaur-shaped. At the moment, players can only access the feature in the beta version of Roblox Studio's avatar editor, and it's unclear when it'll be more widely available.

In the previous tutorial "An introduction to AI in Node.js", we explained two basic approaches for embedding a deep learning model in your Node.js application. In this tutorial, we go a step further and show you how to build and train a simple deep learning model from scratch. Therefore, unlike the previous tutorial, you need a more in-depth understanding of how deep learning models work to get the most benefit from this tutorial. We start with the programming concepts for deep learning and cover two different programming APIs: the high-level Layers API and the low-level Core API. You'll code a simple model to classify clothing items, train it with a small data set, and evaluate the model's accuracy. Then, to illustrate a common practice in deep learning, you'll take your trained model and apply transfer learning to teach the model to classify new items. We also describe how to take a pre-trained model from other sources such as Python and convert it to a format that can be used in JavaScript. So far, we have seen that the actual deep learning model can be hidden in an npm package, loaded from a binary format, or served through a REST API. In these cases, we are simply running an inference on the model, and we don't care how the model was implemented.

Row A - original background, row B - background replaced with a background similar to the one in VITON dataset. We found images of a person who had a similar pose and camera perspective to the training dataset images and saw numerous artifacts present after processing (Row A). However, after removing the unusual background texture and filling the area with the same background color as in the training dataset, the received output quality was improved (although some artifacts were still present). When testing the model using more images, we discovered that the model performed semi-decently on the images similar to the ones from the training distribution and failed completely where the input was distinct enough. You can see the more successful attempts of applying the model and the typical issues we found in Fig 12.

Face and Image Recognition is not only about security and surveillance or controlling the quality of industrial production processes. The technology is proving increasingly impactful to the fashion and beauty industries, generating multiple exciting opportunities for manufacturers and consumers alike. Face and Image recognition being an AI frontrunner in terms of security, agriculture, and industrial QA, the technology's business uses beyond these three realms are still much less known. As a result, many businesses in industries other than security and surveillance, agriculture, and industrial production have barely given any thought to employing Image Recognition as a means of attaining better capabilities to raise their sights and achieve higher levels of quality and profitability. Meanwhile, the Image Recognition- inspired and - enabled opportunities, which have been cropping up of late elsewhere, can barely be ignored and should be taken note of by a much, much wider audience.

In 2002, Target hired statistician Andrew Pole. His job was to use predictive analytics -- a form of statistics that makes predictions by observing data trends -- to help the retail giant market certain products to certain groups of people. Along those lines, Pole's first task was to identify pregnant women -- specifically women in their second trimester. As Target's marketing team explained to him, new parents are extremely valuable customers whose brand loyalty tends to change when they have kids because they purchase things they probably weren't purchasing before -- like diapers, formula, baby clothes, etc. New parents also tend to be physically exhausted and therefore more prone to do all of their shopping at one place.