Google · Agora

Selected work

Representative Papers

Using a market economy to provision compute resources across planet-wide clusters

May 23, 20092009 IEEE International Symposium on Parallel & Distributed Processing

To address resource supply-demand imbalances—manifesting as shortages and surpluses—across globally distributed heterogeneous computing clusters, this paper proposes a resource rationing mechanism grounded in real-world market economics. Methodologically, it introduces a periodic simulated-clock auction framework integrating utilization-driven reserve-price setting, long-term resource quota modeling, and supply-demand equilibrium pricing, enabling dynamic price signals to guide users’ autonomous job placement decisions. Its key contribution lies in being the first to systematically embed microeconomic market mechanisms into large-scale distributed resource allocation, replacing static quota or immediate-scheduling paradigms. Evaluated on the Google experimental market, the mechanism significantly incentivizes user migration toward underutilized clusters: resource utilization variance decreases by 32%, and shortage rate drops by 41%. These results empirically validate that price-based incentives can effectively drive system-level behavioral optimization and achieve global resource equilibrium.

64 citations2 influentialRead paper

Medical Hallucinations in Foundation Models and Their Impact on Healthcare

Feb 26, 2025arXiv.org

Medical foundation models may generate “hallucinations”—factual, logical, or evidence-inconsistent errors—that jeopardize clinical decision-making and patient safety. To address this, we first propose a multidimensional taxonomy of medical hallucinations and establish a real-world, clinician-annotated benchmark dataset derived from authentic clinical cases; we further validate its clinical impact via an international physician survey. Methodologically, we integrate expert annotation, empirical behavioral surveys, and large language model (LLM) evaluation to systematically assess the efficacy of chain-of-thought (CoT) reasoning and retrieval-augmented generation (RAG) in mitigating hallucinations. Results show both techniques significantly reduce hallucination rates, yet residual hallucinations remain clinically hazardous. Building on these findings, we introduce a patient-safety-centered AI governance and ethics framework, offering theoretical foundations and actionable pathways for responsible deployment of medical AI. (149 words)

41 citationsRead paper

Everybody Prune Now: Structured Pruning of LLMs with only Forward Passes

Feb 08, 2024arXiv.org

Existing structured pruning methods for large language models (LLMs) heavily rely on backpropagation, incurring substantial memory and computational overhead. To address this, we propose Bonsai—the first fully backpropagation-free, gradient-agnostic forward-pass pruning method for LLMs. Bonsai estimates module importance via forward perturbation analysis and performs module-level structured pruning without gradient computation. On a single NVIDIA A6000 GPU, Bonsai efficiently prunes the 8B-parameter LLaMA-3 model at 50% sparsity: memory consumption is reduced to one-half to one-third of conventional backward-based methods; pruning speed doubles; inference latency improves by 100%; and accuracy remains state-of-the-art. By eliminating dependence on gradient computation, Bonsai significantly broadens the feasibility of deploying compressed LLMs on resource-constrained hardware.

25 citations4 influentialRead paper

Linear space streaming lower bounds for approximating CSPs

Jun 24, 2021Electron. Colloquium Comput. Complex.

This work investigates the approximability threshold of constraint satisfaction problems (CSPs) in the streaming model. For $n$-variable CSPs over domain ${0,dots,q-1}$ with $O(n)$ constraints, we prove that any algorithm achieving approximation ratio strictly better than the trivial $1/q$ requires $Omega(n)$ space—establishing the first linear-space lower bound for approximation ratios below $1/2$. Methodologically, we extend the Kapralov–Krachun linear lower-bound technique to general CSPs (surpassing prior $Omega(sqrt{n})$ bounds) via modular-$q$ linear equation encoding, communication complexity analysis, and pseudorandom hard-instance reduction. This yields optimal $q^{-(k-1)}$ inapproximability for Max $k$-LIN mod $q$ with $k>2$, $q>2$. Our results uniformly characterize the streaming hardness of broad CSP subclasses: all nontrivial approximation requires essentially linear space, significantly advancing the theoretical understanding of streaming algorithm limitations.

15 citationsRead paper

Enhanced $H$-Consistency Bounds

Jul 18, 2024

Existing H-consistency bounds rely heavily on strong convexity assumptions, limiting their applicability and yielding loose guarantees. Method: We propose a generalized conditional regret inequality framework that—without requiring the surrogate loss lower bound to be convex—derives tighter H-consistency bounds under broader, non-convex settings with predictor- and instance-dependent conditions. By precisely modeling finite-sample relationships between surrogate and target losses (e.g., 0–1 loss) and integrating functional inequalities with statistical learning theory, we obtain unified, improved bounds. Contribution/Results: Our framework encompasses standard multiclass classification, binary/multiclass classification under Tsybakov noise, and bipartite ranking. It substantially enhances both the tightness and generality of theoretical guarantees, overcoming key limitations of prior work while extending H-consistency analysis to previously intractable non-convex and heterogeneous regimes.

12 citations2 influentialRead paper

Recent publications