arXiv subjects

Valentin Danchev

Publications and source records attributed to Valentin Danchev.

5 recordsLinked to original sources

arXiv2025

Towards deployment-centric multimodal AI beyond vision and language

Multimodal artificial intelligence (AI) integrates diverse types of data via machine learning to improve understanding, prediction, and decision-making across disciplines such as healthcare, science, and engineering. However, most multimodal AI advances focus on models for vision and language data, while their deployability remains a key challenge. We advocate a deployment-centric workflow that incorporates deployment constraints early to reduce the likelihood of undeployable solutions, complementing data-centric and model-centric approaches. We also emphasise deeper integration across multiple levels of multimodality and multidisciplinary collaboration to significantly broaden the research scope beyond vision and language. To facilitate this approach, we identify common multimodal-AI-specific challenges shared across disciplines and examine three real-world use cases: pandemic response, self-driving car design, and climate change adaptation, drawing expertise from healthcare, social science, engineering, science, sustainability, and finance. By fostering multidisciplinary dialogue and open research practices, our community can accelerate deployment-centric development for broad societal impact.

cs.AI↗

arXiv2022

Data Governance in the Age of Large-Scale Data-Driven Language Technology

The recent emergence and adoption of Machine Learning technology, and specifically of Large Language Models, has drawn attention to the need for systematic and transparent management of language data. This work proposes an approach to global language data governance that attempts to organize data management amongst stakeholders, values, and rights. Our proposal is informed by prior work on distributed governance that accounts for human values and grounded by an international research collaboration that brings together researchers and practitioners from 60 countries. The framework we present is a multi-party international governance structure focused on language data, and incorporating technical and organizational tools needed to support its work.

cs.CY↗

arXiv2020

Migration Networks: Applications of Network Analysis to Macroscale Migration Patterns

An emerging area of research is the study of macroscale migration patterns as a network of nodes that represent places (e.g., countries, cities, and rural areas) and edges that encode migration ties that connect those places. In this chapter, we first review advances in the study of migration networks and recent work that has employed network analysis to examine such networks at different geographical scales. In our discussion, we focus in particular on global scale migration networks. We then propose ways to leverage network analysis in concert with digital technologies and online geolocated data to examine the structure and dynamics of migration networks. The implementation of such approaches for studying migration networks faces many challenges, including ethical ones, methodological ones, socio-technological ones (e.g., data availability and reuse), and research reproducibility. We detail these challenges, and we then consider possible ways of linking digital geolocated data to administrative and survey data as a way of harnessing new technologies to construct increasingly realistic migration networks (e.g., using multiplex networks). We also briefly discuss new methods (e.g., multilayer network analysis) in network analysis and adjacent fields (e.g., machine learning) that can help advance understanding of macroscale patterns of migration.

cs.SI↗

arXiv2018

Centralized "big science" communities more likely generate non-replicable results

Growing concern that most published results, including those widely agreed upon, may be false are rarely examined against rapidly expanding research production. Replications have only occurred on small scales due to prohibitive expense and limited professional incentive. We introduce a novel, high-throughput replication strategy aligning 51,292 published claims about drug-gene interactions with high-throughput experiments performed through the NIH LINCS L1000 program. We show (1) that unique claims replicate 19% more frequently than at random, while those widely agreed upon replicate 45% more frequently, manifesting collective correction mechanisms in science; but (2) centralized scientific communities perpetuate claims that are less likely to replicate even if widely agreed upon, demonstrating how centralized, overlapping collaborations weaken collective understanding. Decentralized research communities involve more independent teams and use more diverse methodologies, generating the most robust, replicable results. Our findings highlight the importance of science policies that foster decentralized collaboration to promote robust scientific advance.

cs.SI↗

arXiv2016

Heterogeneity of Global and Local Connectivity in Spatial Network Structures of World Migration

We examine world migration as a social-spatial network of countries connected via movements of people. We assess how multilateral migratory relationships at global, regional, and local scales coexist ("glocalization"), divide ("polarization"), or form an interconnected global system ("globalization"). To do this, we decompose the world migration network (WMN) into communities---sets of countries with denser than expected migration connections---and characterize their pattern of local (i.e., intracommunity) and global (i.e., intercommunity) connectivity. We distinguish community signatures---"cave", "biregional", and "bridging"---with distinct migration patterns, spatial network structures, temporal dynamics, and underlying antecedents. Cave communities are tightly-knit, enduring structures that tend to channel local migration between contiguous countries; biregional communities are likely to merge migration between two distinct geographic regions (e.g., North Africa and Europe); and bridging communities have hub-and-spoke structures that tend to emerge dynamically from globe-spanning movements. We find that world migration is neither globally interconnected nor reproduces the geographic boundaries as drawn on a world map but involves a heterogeneous interplay of global and local tendencies in different network regions. We discuss the implications of our results for the understating of variability in today's transnational mobility patterns and migration opportunities across the globe.

physics.soc-ph↗