🏢 Google Research

New algorithm efficiently solves linear regression with heterogeneous data batches, handling diverse input distributions and achieving high accuracy with fewer samples.

This paper shows how noisy predictions about optimal solutions can improve approximation algorithms for NP-hard problems like MAX-CUT, exceeding classical hardness bounds.

Learn optimal LP values faster with a novel neural network method!

IllumiNeRF: Relightable 3D reconstruction without inverse rendering using image diffusion and NeRF.

This paper presents a novel approach for embedding supervised models in hyperbolic space, linking loss functions to hyperbolic distances and introducing monotonic decision trees for unambiguous visual…

Boosting, traditionally limited by assumptions about loss functions, is proven in this paper to efficiently optimize any loss function regardless of differentiability or convexity.

Generative Forests (GFs) revolutionize tabular data generation with a novel forest-based model and a simple boosting algorithm offering strong convergence guarantees, significantly outperforming curre…

Long-video masked autoencoders (LVMAE) achieve state-of-the-art performance by using an adaptive masking strategy that prioritizes important video tokens, enabling efficient training on 128 frames.

EAGLE: Guiding LLMs using latent embeddings for controlled text generation.

First-price auction efficiency in autobidding plummets to 45.7% with mixed bidders, but machine-learned advice restores optimality.

DynaMITE-RL: A new meta-RL approach masters environments with evolving latent states by cleverly modeling episode sessions and refining existing meta-RL techniques.

This paper presents new, nearly optimal differentially private algorithms for handling sparse gradients, significantly improving efficiency and scalability in large embedding models.

In symmetric zero-sum games, no-swap-regret dynamics guarantee strong convergence to Nash Equilibrium under symmetric initial conditions, but this advantage disappears when constraints are relaxed.

Repeated contracts with learning agents are optimized by a simple dynamic contract: initially linear, then switching to zero-cost, causing the agent’s actions to ‘free-fall’ and yield non-zero rewards…

This paper proposes cardinality-aware top-k classification, improving accuracy and efficiency by dynamically adjusting prediction set sizes.

New algorithms conquer adversarial low-rank MDPs, improving regret bounds for unknown transitions and bandit feedback.

More sophisticated autobidders surprisingly worsen online auction efficiency; a fine-grained analysis reveals that less powerful, uniform bidders lead to better market outcomes.

New metrics audit label privatization, revealing differentially private schemes often outperform heuristic methods in the privacy-utility tradeoff.

Random-order stream data necessitates efficient top eigenvector approximation; this paper presents novel algorithms with improved space complexity, achieving near-optimal bounds.