We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to asemantic cluster formed by local decoder feature points with semantic similarity. By rethinking point affiliation, we present a generic formulation for generating upsampling kernels. The kernels encourage notonly semantic smoothness butalsoboundary sharpness intheupsampled feature maps. Such properties are particularly useful for some dense prediction tasks such as semantic segmentation. The key idea of our formulation istogenerate similarity-awarekernels bycomparing thesimilarity between each encoder feature point and the spatially associated local region of decoder features.

Uncertainty estimation is essential to make neural networks trustworthy in real-world applications. Extensive research efforts have been made to quantify and reduce predictive uncertainty. However, most existing works are designed for unimodal data, whereas multi-view uncertainty estimation has not been sufficiently investigated. Therefore, we propose a new multi-view classification framework for better uncertainty estimation and out-of-domain sample detection, where we associate each view with an uncertainty-aware classifier and combine the predictions of all the views in a principled way.

We train a model to generate images from multimodal prompts of interleaved text and images such as "a man and his dog in an animated style." We bootstrap a multimodal dataset by extracting semantically meaningful image crops corresponding to words in the image captions of synthetically generated and publicly available text-image data. Our model, MUMU, is composed of a vision-language model encoder with a diffusion decoder and is trained on a single 8xH100 GPU node. Despite being only trained on crops from the same image, MUMU learns to compose inputs from different images into a coherent output. For example, an input of a realistic person and a cartoon will output the same person in the cartoon style, and an input of a standing subject and a scooter will output the subject riding the scooter. As a result, our model generalizes to tasks such as style transfer and character consistency. Our results show the promise of using multimodal models as general purpose controllers for image generation.

This article was written by Swami Chandrasekaran. Click on picture to zoom in. When I wrote my blog post Becoming a Data Scientist-- Curriculum via Metromap, little did I know that it will receive a rousing feedback. Over years a lot of people reached out to me with very kind words and how they use it as a guide in their data scientist journey. Also, many who sought permission to use the Metromap picture in their presentations as well as a few universities that also reached out to use it as part of their syllabus.

AI Artwork is a process of Machine Learning to create things. Creating artwork starts with deciding what you want to make, whether it's a painting, a drawing, or a sculpture. Once you have an idea in mind, you start the creation process using generative techniques.

The latest research, conducted by Samsung's AI Centre, looked specifically at making a system that can recreate lifelike motion from only one single frame of a person's face. This basically means, using a still image of either a painting or just a normal photograph to make it appear as if it is speaking.

In this article, we investigate the possibility of cross-language communication using a synergy of words and pictures on mobile devices. On the one hand, communicating with only pictures is in itself a very powerful strategy, but is limited in expressiveness. On the other hand, words can express everything you could wish to say, but they are cumbersome to work with on mobile devices and need to be translated in order for their meaning to be understood. Automatic translations can contain errors that pervert the communication process, and this may undermine the users' confidence when expressing themselves across language barriers. Our idea is to create a user interface for cross-language communication that uses pictures as the primary mode of input, and words to express the detailed meaning.

The Turing test, as originally conceived, focused on language and reasoning; problems of perception and action were conspicuously absent. To serve as a benchmark for motivating and monitoring progress in AI research, this article proposes an extension to that original proposal that incorporates all four of these aspects of intelligence. Some initial suggestions are made regarding how best to structure such a test and how to measure progress. The proposed test also provides an opportunity to bring these four important areas of AI research back into sync after each has regrettably diverged into a fairly independent area of research of its own. He observed, however, that such a goal was somewhat ill-defined: how was one to conclude whether or not a machine was thinking (like a human)?

IAAI seeks out applications of artificial intelligence that either demonstrate new technology or use previously known technology in innovative ways. IAAI particularly seeks out examples of deployments of AI technology that tackle the problems of demonstrating value and planning for long-term deployment. The five articles we have selected for this special issue are extended versions of papers that appeared in the conference. Two of the articles are deployed applications that have already demonstrated practical value. The remaining three articles are particularly innovative emerging applications.

Today, as the world moves into one in which everyone owns at least one mobile device, be it a smartphone, a tablet, or other handheld device, applications on the devices are increasingly more intelligent as well. We will see more and more applications of AI on the mobile devices. This special issue of AI Magazine is devoted to some exemplary works of AI on mobile devices. We include four works that range from mobile activity recognition and air-quality detection to machine translation and image compression. These works were chosen from a variety of sources, including the International Joint Conference on Artificial Intelligence 2011 Special Track on Integrated and Embedded AI Systems, held in Barcelona, Spain, in July 2011. In "User-Centric Indoor Air-Quality Monitoring on Mobile Devices," written by Yifei Jiang, Kun Li, Ricardo Piedrahita, Yun Xiang, Lei Tian, Omkar Mansata, Qin Lv, Robert P. Dick, Michael Hannigan, and Li Shang, the authors develop a novel and important technique for portable indoor air quality (IAQ) detec-