Institution profile

Stockholm University

Academic institutioneurope · se
Official website
Research library173linked papers
Opportunities0open roles
Selected work

Representative Papers

Learning conformational ensembles of proteins based on backbone geometry

Feb 19, 2025arXiv.org

Existing protein conformational sampling methods—relying either on evolutionary information or pretrained folding models—suffer from limited applicability, low efficiency, and potential biases. To address these limitations, we propose BBFlow, the first flow-matching generative model that operates exclusively on backbone geometric structure, requiring neither evolutionary sequence information nor pretrained models, and directly learns a conformational ensemble consistent with the Boltzmann distribution from scratch. BBFlow innovatively employs equilibrium backbone geometry both to condition the vector field and to define a learnable SE(3)-equivariant prior distribution, enabling robust modeling of multi-chain proteins and de novo design. Compared to state-of-the-art methods, BBFlow achieves orders-of-magnitude faster training (converging in GPU-days) and significantly accelerated inference, while maintaining competitive performance on both native protein reconstruction and de novo design benchmarks.

2 citations1 influentialRead paper

AI-Enhanced Business Process Automation: A Case Study in the Insurance Domain Using Object-Centric Process Mining

Apr 24, 2025

Manual identification of claim components in insurance claims processing creates a scalability bottleneck. Method: This study deploys large language models (LLMs) in a real production environment to automate knowledge-intensive tasks and introduces object-centric process mining (OCPM) for the first time to dynamically model AI-augmented process evolution. The approach integrates LLM-based reasoning, event log analysis, and OCPM modeling to quantitatively assess performance changes. Results: LLM deployment improves component identification efficiency by 3.2× and doubles daily throughput. OCPM uncovers three types of AI-induced latent rework paths, as well as novel process couplings and anomalous patterns. This work empirically validates OCPM’s capability to characterize AI-driven process evolution and establishes a reusable, evidence-based evaluation framework for human-AI collaboration optimization in knowledge-intensive domains.

1 citationsRead paper

Better Together? A Field Experiment on Human-Algorithm Interaction in Child Protection

Feb 12, 2025

This study addresses inefficiency and inequity in Child Protective Services (CPS) investigation assignment. We conduct a randomized controlled trial (RCT) with frontline caseworkers to evaluate an algorithm-augmented decision-support system. Our key contribution is the empirical identification of *human–algorithm complementarity*: caseworkers deliberately intensify scrutiny of high-risk children whom the algorithm classifies as low-risk—particularly reducing over-investigation of Black children. Causal inference and counterfactual simulations demonstrate that this collaborative dynamic lowers child abuse–related hospitalizations and reduces child injury rates by 29%. Moreover, caseworkers significantly increase their review of supplementary information prompted by algorithmic alerts. To our knowledge, this is the first field study in public service delivery to rigorously validate human–algorithm complementarity while simultaneously improving operational efficiency, procedural fairness, and substantive child welfare outcomes.

1 citationsRead paper

Comparing Semantic Frameworks for Dependently-Sorted Algebraic Theories

Dec 27, 2024Asian Symposium on Programming Languages and Systems

Existing work lacks a systematic characterization and comparative analysis of the relationships among various categorical semantic models—such as fibrations, sheaves, and extensions of homotopy type theory—for dependently sorted algebraic theories. Method: This paper establishes, for the first time, a semantic taxonomy for dependently sorted algebraic theories, introducing cross-framework expressivity hierarchies and structure-preserving translation mechanisms. Leveraging category theory, semantics of dependent types, and fibration-based methods, we rigorously determine the relative expressivity ordering among three mainstream semantic frameworks. Contribution/Results: We precisely characterize equivalence boundaries and prove fundamental intranslatability results between certain models. Our framework unifies the treatment of type dependency and algebraic structure, yielding a comparable and transferable semantic foundation for higher-order formal verification.

1 citationsRead paper

Automating Boundary Filling in Cubical Agda

Feb 19, 2024International Conference on Formal Structures for Computation and Deduction

This work addresses the challenge of automating proofs involving higher-dimensional homotopical structures in cubical type theory, focusing on boundary filling—the core task underlying both contortion solving (single-cube filling) and Kan solving (multi-cube gluing). Methodologically, it models contortion as a poset-mapping problem and introduces, for the first time, a constraint satisfaction programming (CSP) framework to solve Kan filling. To mitigate combinatorial explosion in high dimensions and undecidability inherent in the theory, it designs lightweight heuristic algorithms. The prototype solver, implemented in Haskell, integrates poset-based modeling, CSP solving, and an interface to Cubical Agda. It successfully automates the verification of the Eckmann–Hilton theorem and demonstrates efficiency across multiple benchmark suites. This work establishes a novel paradigm and delivers key technical foundations for practical proof automation in cubical type theory.

1 citationsRead paper
Recent publications

Latest Papers