We study how inherent randomness in the training process—where each sample (or client in federated learning) contributes only to a randomly selected portion of training—can be leveraged for privacy amplification. This includes (1) data partitioning, where a sample participates in only a subset of training iterations, and (2) model partitioning, where a sample updates only a subset of the model parameters. We apply our framework to model parallelism in federated learning, where each client updates a randomly selected subnetwork to reduce memory and computational overhead, and show that existing methods, e.g. model splitting or dropout, provide a significant privacy amplification gain not captured by previous privacy analysis techniques. Additionally, we introduce balanced iteration subsampling, a new data partitioning method where each sample (or client) participates in a fixed number of training iterations. We show that in certain regimes, this method yields stronger privacy amplification than Poisson (i.i.d.) sampling of data (or clients). Our results demonstrate that randomness in the training process, which is structured rather than i.i.d. and interacts with data in complex ways, can be systematically leveraged for nontrivial privacy amplification.

This paper addresses a new dimension in the evaluation of creative ideas. This is based on the observation that some ideas seem to trigger or lead to more ideas than other ideas. We study the principle of idea richness --the potential for ideas to trigger more ideas. We hypothesize a correlation between creative value as ascribed by experts and idea richness as captured empirically from idea sharing in online communities. The paper discusses ongoing work and addresses the implications of the possible outcomes of this research.

This paper discusses a theory for the foundation of design for pedagogy based on the principles of pattern languages. We develop a pattern using a variation of the Alexandrian pattern structure to embed pedagogy at the core of the design of e-learning. A pattern dealing with the organization of an online discussion group based on the principles of constructivism and experiential learning is produced to illustrate the application of the theory.

This article deals with the organization and management of the computer files handled by mechanical engineers on their personal computers. In engineering organizations, a wide variety of electronic files (documents) are necessary to support both business processes and the activities of design and manufacture. Whilst a large number of files and hence information is formally archived, a significant amount of additional information and knowledge resides in electronic files on personal computers. The widespread use of these personal information stores means that all information is retained. However, its reuse is problematic for all but the individual as a result of the naming and organization of the files. To begin to address this issue, a study of the use and current practices for managing personal electronic files is described. The study considers the fundamental classes of files handled by engineers and analyses the organization of these files across the personal computers of 40 participants. The study involves a questionnaire and an electronic audit. The results of these qualitative and quantitative elements are used to elicit an understanding of the practices and requirements of engineers for managing personal electronic files. A potential scheme for naming and organizing personal electronic files is discussed as one possible way to satisfy these requirements. The aim of the scheme is to balance the personal nature of data storage with the need for personal records to be shared with others to support knowledge reuse in engineering organizations. Although this article is concerned with mechanical engineers, the issues dealt with are relevant to knowledge-based industries and, in particular, teams of knowledge workers.

The paper discusses the digital artwork series Impossible Geographies, works that weave dynamic cartographies of invisible, fragile and hybrid spaces. Impossible Geographies 01: Memory invokes the memory space of a gallery, producing cracks through which past events seep into the physical present. Net_Dérive negotiates the space between a gallery and its urban neighborhood, creating hybrid narratives along the grooves left by participants equipped with mobile phones. The playground of Impossible Geographies 02: Urban Fiction spreads across the city and beyond, mixing and shifting multiple urban geographies as participants sweep along the urban fabric with advanced mobile phones. The conceptual and technological methods of 'lens-making' and 'spacing,' developed in these artworks, explore the subjective, hybrid, and migrational nature of the geographies we belonging to. By re-sculpting the fluid nature of the urban fabric the works transform the everyday urbanscape into impossible imaginary geographies.