In recent years, federated learning has been embraced as an approach for bringing about collaboration across large populations of learning agents. However, little is known about how collaboration protocols should take agents’ incentives into account when allocating individual resources for communal learning in order to maintain such collaborations. Inspired by game theoretic notions, this paper introduces a framework for incentive-aware learning and data sharing in federated learning. Our stable and envy-free equilibria capture notions of collaboration in the presence of agents interested in meeting their learning objectives while keeping their own sample collection burden low. For example, in an envy-free equilibrium, no agent would wish to swap their sampling burden with any other agent and in a stable equilibrium, no agent would wish to unilaterally reduce their sampling burden. In addition to formalizing this framework, our contributions include characterizing the structural properties of such equilibria, proving when they exist, and showing how they can be computed. Furthermore, we compare the sample complexity of incentive-aware collaboration with that of optimal collaboration when one ignores agents’ incentives.

Distribution shifts—where the training distribution differs from the test distribution—can substantially degrade the accuracy of machine learning (ML) systems deployed in the wild. Despite their ubiquity in the real-world deployments, these distribution shifts are under-represented in the datasets widely used in the ML community today. To address this gap, we present WILDS, a curated benchmark of 10 datasets reflecting a diverse range of distribution shifts that naturally arise in real-world applications, such as shifts across hospitals for tumor identification; across camera traps for wildlife monitoring; and across time and location in satellite imaging and poverty mapping. On each dataset, we show that standard training yields substantially lower out-of-distribution than in-distribution performance. This gap remains even with models trained by existing methods for tackling distribution shifts, underscoring the need for new methods for training models that are more robust to the types of distribution shifts that arise in practice. To facilitate method development, we provide an open-source package that automates dataset loading, contains default model architectures and hyperparameters, and standardizes evaluations. The full paper, code, and leaderboards are available at https://wilds.stanford.edu.

Motivated by the need to audit complex and black box models, there has been extensive research on quantifying how data features influence model predictions. Feature influence can be direct (a direct influence on model outcomes) and indirect (model outcomes are influenced via proxy features). Feature influence can also be expressed in aggregate over the training or test data or locally with respect to a single point. Current research has typically focused on one of each of these dimensions. In this paper, we develop disentangled influence audits, a procedure to audit the indirect influence of features. Specifically, we show that disentangled representations provide a mechanism to identify proxy features in the dataset, while allowing an explicit computation of feature influence on either individual outcomes or aggregate-level outcomes. We show through both theory and experiments that disentangled influence audits can both detect proxy features and show, for each individual or in aggregate, which of these proxy features affects the classifier being audited the most. In this respect, our method is more powerful than existing methods for ascertaining feature influence.

No abstract available.

MediaView is a computer program that provides a generic infrastructure for authoring and interacting with multimedia documents. Among its applications is the ability to furnish a user with a comprehensive environment for analysis and visualization. With this program the user can produce a document that contains mathematics, datasets and associated visualizations. From the dataset or embedded mathematics animated sequences can be produced in situ. Equations that appear in a document have a backing format that is compatible with the Mathematica language. Thus, by clicking on an equation, its semantics are conveyed to Mathematica, where the user can perform a variety of symbolic and numerical operations. Since the document is all digital, it can be shared on a local network or mailed electronically to a distant site. Animations and any other substructures of the document persist through the mailing process and can be awakened at the destination by the recipient.>

A generalized high level query language has been developed for use with a network data base, as defined by the CODASYL DBTG report. Of particular interest is the capability of manipulating graphical entities with this language. The system was developed to supportan offshore oil lease information project where one must formulate queries pertaining to proximity of graphical entities. To facilitate such requests new relational operators were defined and implemented. Thus compound Boolean expressions can be formed which involve graphical as well as non-graphical criteria. Since one goal of the information system is to produce selectively colored maps via an FR80 microfilm recorder, the query language syntax provides the user with a comprehensive thematic cartography capability. The query language is designed to isolate the user from the underlying network structure of the data base. In order to accomplish this generalized pathfinding and accessing algorithms had to be developed. Pathfinding is facilitated through an adjacency matrix involving all record types in the data base, while accession is controlled by examining the local owner-member relation for current record types. Besides querying, the system provides the usual complement of data base management system functions, i.e. deletion, updating, reporting, etc.