36 research outputs found
Faster Rates for Compressed Federated Learning with Client-Variance Reduction
No full textDue to the communication bottleneck in distributed and federated learning
applications, algorithms using communication compression have attracted
significant attention and are widely used in practice. Moreover, the huge
number, high heterogeneity and limited availability of clients result in high
client-variance. This paper addresses these two issues together by proposing
compressed and client-variance reduced methods COFIG and FRECON. We prove an
bound on the number of communication rounds of COFIG in the nonconvex setting,
where is the total number of clients, is the number of clients
participating in each round, is the convergence error, and
is the variance parameter associated with the compression operator. In case of
FRECON, we prove an bound on the
number of communication rounds. In the convex setting, COFIG converges within
communication rounds, which, to the
best of our knowledge, is also the first convergence result for compression
schemes that do not communicate with all the clients in each round. We stress
that neither COFIG nor FRECON needs to communicate with all the clients, and
they enjoy the first or faster convergence results for convex and nonconvex
federated learning in the regimes considered. Experimental results point to an
empirical superiority of COFIG and FRECON over existing baselines.Comment: Accepted by SIAM Journal on Mathematics of Data Science (SIMODS
Enabling pervasive computing with smart phones
No full textThe authors discuss their experience with a number of mobile telephony projects carried out in the context of the European Union Information Society Technologies research program, which aims to develop mobile information services. They identify areas where use of smart phones can enable pervasive computing and offer practical advice in terms of lessons learned. To this end, they first look at the mobile telephone as * the end point of a mobile information service,* the control device for ubiquitous systems management and configuration,* the networking hub for personal and body area networks, and* identification tokens.They conclude with a discussion of business and practical issues that play a significant role in deploying research systems in realistic situations
DIVE: a generic tool for the deployment of shared virtual environments
No full textIn this paper we present the DIVE system (Distributed Interactive Virtual Environment), an architecture and a programming toolkit for the realisation and implementation of wide-area, Internet-based multi-user virtual environments. Along the years, DIVE has evolved into a generic tool that supports a wide range of applications and situations. This paper focuses on the networking and architectural aspects that allow the deployment of multi-user virtual environments on the Internet. Additionally, this paper presents the palette of programming interfaces and techniques that are offered by the system. Our experience has proven that its ability to mix interfaces made it a system of choice for the implementation of distributed applications. We conclude with a selection of applications and how they make use of the networking and programming features of the system
Teaching Virtual Characters to use Body Language
No full textNon-verbal communication, or “body language”, is a critical component in constructing believable virtual characters. Most often, body language is implemented by a set of ad-hoc rules.We propose a new method for authors to specify and refine their character’s body-language responses. Using our method, the author watches the character acting in a situation, and provides simple feedback on-line. The character then learns to use its body language to maximize the rewards, based on a reinforcement learning algorithm
Virtually Telling Robots What to Do
No full textThis paper describes a distributed virtual environment application that combines robotics in both virtual and real worlds. The motivation for this system comes from research in virtual and augmented reality, autonomous robotics and computer vision. The system combines graphical immersiveenvironments with the live video from a robot working in a real environment. The worlds are synchronized and updated based on both the operator selections, commands and robot actions. This system allows the user to have apowerful tool to create and control autonomous robots, thus making possible the realization of single and multiple autonomous robot applications
THE STATE IN - INVESTIGATING A POSSIBLE GATEWAY TO CORE NON-PENETRATING RYDBERG STATES
No full textAuthor Institution: Department of Physics, United States Military Academy; Department of Physics, Temple University; Department of Chemistry, University of Illinois at Chicago; Key Lab Atom and Molecular Nanoscience, Tsinghua University; Laboratoire de Physique des Atomes, Lasers, Mol\'{e}cules et Surfaces, (PALMES), CNRS et Universit\'{e}; Laboratoire de Spectrom\'etrie Ionique et Moleculaire (L.A.S.I.M), CNRS et Universit\'e Lyon (UMR5579)Core non-penetrating Rydberg states can give useful information on the electronic structure of the ion core; however, core non-penetrating states are difficult to observe since these states hardly penetrate the more accessible ion core and the electronic angular momentum quantum number, l, is large, for the core non-penetrating states thus the transition dipole moment to the core non-penetrating states is small. The core penetrating state (atomic limit: ) and the core non-penetrating state (atomic limit: ) perturb each other since they have the same symmetry and overlapping energy states thus creating the possibility of a gateway to other core non-penetrating states
The extended museum visit: documenting and exhibiting post-visit experiences
No full textDuring the last couple of decades, a growing body of research has provided insights into the complex processes of learning that take place in museums. Interestingly, museum-related learning is not limited to the actual visit: what takes place before and afterwards has a profound effect on the learning outcome. The study presented in this paper focuses on the post-visit aspects of the learning process. Previous research shows that visitors make connections between their experiences in the museum and experiences that happen after the visit. Sometimes these connections can occur weeks or months (or even years) after the visit, depending on when the visitor happens upon a situation that allows the connection to be made. Documenting these events is obviously quite difficult. Even though it is possible to re-establish contact with visitors after a few weeks or months (e.g., through telephone or e-mail), the information obtained is not in situ. The goal of the present study is to attempt to acquire and analyze more data from these in situ situations, and to re-present the data in an exhibition. To this end, we have designed a system that allows visitors to send images and text messages to a central server through e-mail, SMS or MMS. The data from the server can then be visualized as a weblog (blog) or in some other suitable form. We collaborated with the Museum of Science and Technology in Stockholm, Sweden. A large part of the Museum's Science Centre is devoted to five mechanical principles: the screw, the plane slope, the lever, the wheel and the wedge. We have designed an exhibit that utilizes our system to present messages (images and text) from visitors on the subject of the five principles in the science centre itself. It is also possible to access the messages through a public Web page. From the Museum's perspective, our exhibit not only provides new opportunities for documenting post-visit learning processes, but also has the potential to provide new forms of evaluation data that might be difficult to obtain through other means. Furthermore, it allows Museum visitors to extend the original scope of the mechanical principles exhibition by allowing them to provide their own content (and reflections upon the existing content), a re-occurring theme in many recent technology-based exhibits. The paper provides a description of our system, the exhibit we have built, how the exhibit is managed, and how it has worked in practice.</p
