arXiv subjects

Brian Y. C. Leung

Publications and source records attributed to Brian Y. C. Leung.

3 recordsLinked to original sources

arXiv2026

Learned Cross-Task Relationships in Multi-Task Models

We propose a framework that learns cross-task relationships in multi-task models by approximating the joint distribution of task labels through targeted pairwise relationships. This approach improves performance via transfer learning and enhances information extraction without the intractable complexity of modeling the full joint space. Although our framework applies to any multi-task system, we demonstrate its efficacy within YouTube's production recommendation systems. Experiments across the Notifications, Homepage, and Watch Next surfaces show improvements in both accuracy and user satisfaction metrics. Finally, we propose a workflow template to facilitate broader future implementation.

cs.AI↗

arXiv2026

Bridging the Gap Between Natural Language and Market Dynamics via High-Dimensional Representation Learning

Traditional multi-modal financial forecasting often relies on scalar sentiment scores, which fail to capture the nuances of financial news. To address this information loss, this paper explores high-dimensional representation learning by replacing discrete polarity ratings with dense FinBERT embeddings within a Transformer-based forecasting architecture. We benchmarked various embedding strategies on the FNSPID dataset, including raw embeddings, attention-weighted aggregation, and a custom Siamese network. While the attention-based mechanism struggled with the low signal-to-noise ratio typical of financial data, the integration of Siamese-optimized embeddings outperformed both the scalar baseline and raw embedding approaches, demonstrating that preserving high-dimensional narrative context yields improved predictive accuracy for short-term stock price movements.

cs.LG↗

arXiv2013

Isotropic Band Gaps and Freeform Waveguides Observed in Hyperuniform Disordered Photonic Solids

Recently, disordered photonic media and random textured surfaces have attracted increasing attention as strong light diffusers with broadband and wide-angle properties. We report the first experimental realization of an isotropic complete photonic band gap (PBG) in a two-dimensional (2D) disordered dielectric structure. This structure is designed by a constrained-optimization method, which combines advantages of both isotropy due to disorder and controlled scattering properties due to low density fluctuations (hyperuniformity) and uniform local topology. Our experiments use a modular design composed of Al2O3 walls and cylinders arranged in a hyperuniform disordered network. We observe a complete PBG in the microwave region, in good agreement with theoretical simulations, and show that the intrinsic isotropy of this novel class of PBG materials enables remarkable design freedom, including the realization of waveguides with arbitrary bending angles impossible in photonic crystals. This first experimental verification of a complete PBG and realization of functional defects in this new class of materials demonstrates their potential as building blocks for precise manipulation of photons in planar optical micro-circuits and has implications for disordered acoustic and electronic bandgap materials.

cond-mat.mes-hall↗