arXiv subjects

James Li

Publications and source records attributed to James Li.

13 recordsLinked to original sources

arXiv2026

UniPinRec: Unifying Generative Retrieval and Ranking at Pinterest Scale

Modern recommendation systems predominantly train retrieval and ranking as separate models despite both increasingly relying on large transformers encoding the same user behavior data, duplicating parameters, compute, and serving cost. Prior work unifies the model architecture but not the full pipeline: input formats, training procedures, and serving stacks remain fragmented across stages. We present UniPinRec, which achieves full-stack unification of retrieval and ranking at Pinterest: one input format, one model, one training stage, deployed within existing serving infrastructure. A shared transformer encodes the user action sequence into candidate-independent representations that branch into retrieval (ANN dot-product) and ranking (cross-attention) via task-specific heads. Three ideas make this work: (1) Masked Action Modeling (MAM) eliminates interleaving, enabling weight sharing without doubling context length; (2) Blended training examples pair action sequences with feedview impression slates to satisfy both objectives jointly; (3) Cross-stage KV cache sharing reuses user-history computation from retrieval for ranking, reducing total FLOPs versus serving two independent models. Deployed in the Pinterest core surfaces, UniPinRec delivers approximately +1% online engagement lift while cutting end-to-end serving latency by 11.1% and lifting QPS by 63.6%. To our knowledge, this is the first full-stack unification of retrieval and ranking, covering inputs, model, training and serving, deployed in a production recommendation system.

cs.IR↗

arXiv2026

Motubrain: An Advanced World Action Model for Robot Control

Vision-Language-Action (VLA) models generalize semantically well but often lack fine-grained modeling of world dynamics. We present Motubrain, a unified World Action Model that jointly models video and action under a UniDiffuser formulation with a three-stream Mixture-of-Transformers architecture. A single model supports policy learning, world modeling, video generation, inverse dynamics, and joint video-action prediction, while scaling to heterogeneous multimodal data such as video-only, task-agnostic, and cross-embodiment robot data. Building on Motus, Motubrain further introduces unified multiview modeling, an independent text stream for stronger language-action coupling, a shared cross-embodiment action representation, and an efficient post-training and deployment recipe for long-horizon real-world control. Our inference stack combines step reduction, compilation, FP8 quantization, DiT caching, V2A-style action-only inference, and real-time chunked closed-loop execution, achieving over 50x speedup over a naive baseline and up to 11 Hz inference. Experimentally, Motubrain achieves 95.8% and 96.1% average success on RoboTwin 2.0 under clean and randomized settings, respectively, attains the strongest reported EWMScore in our WorldArena comparison, and adapts to new humanoid embodiments with only 50--100 trajectories. These results show that unified world action models can scale in generality, predictive accuracy, and real-world deployability.

cs.RO↗

arXiv2026

Quantization Robustness of Monotone Operator Equilibrium Networks

Monotone operator equilibrium networks are implicit-layer models whose output is the unique equilibrium of a monotone operator, guaranteeing existence, uniqueness, and convergence. When deployed on low-precision hardware, weights are quantized, potentially destroying these guarantees. We analyze weight quantization as a spectral perturbation of the underlying monotone inclusion. Convergence of the quantized solver is guaranteed whenever the spectral-norm weight perturbation is smaller than the monotonicity margin; the displacement between quantized and full-precision equilibria is bounded in terms of the perturbation size and margin; and a condition number characterizing the ratio of the operator norm to the margin links quantization precision to forward error. MNIST experiments confirm a phase transition at the predicted threshold: three- and four-bit post-training quantization diverge, while five-bit and above converge. The backward-pass guarantee enables quantization-aware training, which recovers provable convergence at four bits.

math.OC↗

arXiv2025

Synergizing Implicit and Explicit User Interests: A Multi-Embedding Retrieval Framework at Pinterest

Industrial recommendation systems are typically composed of multiple stages, including retrieval, ranking, and blending. The retrieval stage plays a critical role in generating a high-recall set of candidate items that covers a wide range of diverse user interests. Effectively covering the diverse and long-tail user interests within this stage poses a significant challenge: traditional two-tower models struggle in this regard due to limited user-item feature interaction and often bias towards top use cases. To address these issues, we propose a novel multi-embedding retrieval framework designed to enhance user interest representation by generating multiple user embeddings conditioned on both implicit and explicit user interests. Implicit interests are captured from user history through a Differentiable Clustering Module (DCM), whereas explicit interests, such as topics that the user has followed, are modeled via Conditional Retrieval (CR). These methodologies represent a form of conditioned user representation learning that involves condition representation construction and associating the target item with the relevant conditions. Synergizing implicit and explicit user interests serves as a complementary approach to achieve more effective and comprehensive candidate retrieval as they benefit on different user segments and extract conditions from different but supplementary sources. Extensive experiments and A/B testing reveal significant improvements in user engagements and feed diversity metrics. Our proposed framework has been successfully deployed on Pinterest home feed.

cs.IR↗

arXiv2024

Total Outcome Logic: Unified Reasoning for a Taxonomy of Program Logics

While there is a long tradition of reasoning about (non)termination in program analysis, specialized logics are typically needed to give different termination criteria. This includes partial correctness, where termination is not guaranteed, and total correctness, where it is guaranteed. We present Total Outcome Logic (TOL), a single logic which can express the full spectrum of termination conditions and program properties offered by the aforementioned logics. TOL extends (non)termination and (in)correctness reasoning across different kinds of branching effects, so that a single metatheory powers this reasoning in different kinds of programs, including nondeterministic and probabilistic. We also show that TOL subsumes several recently created taxonomies of (in)correctness logics, so that many different kinds of properties can be proven with a single unified theory.

cs.LO↗

arXiv2022

OSC Community Lab: The Integration Test Bed for O-RAN Software Community

O-RAN Software Community (OSC) is an open-source project collaborated by O-RAN Alliance and Linux Foundation, aiming to develop reference software components based on 3GPP and O-RAN Alliance specifications. The OSC has twelve projects. Among them, the Integration and Testing (INT) project is responsible for testing the requirements documented in each release for end-to-end and use case testing. Three OSC Community Laboratories were built to speed up the integration and interoperability testing among different projects. This paper summarizes the software components developed by OSC projects and the status of the three OSC Community Laboratories. The activities of each laboratory, how the community collaborates, and the challenges we encountered along the way were elaborated.

cs.NI↗

arXiv2022

Venus as an Exoplanet: I. An Initial Exploration of the 3-D Energy Balance for a CO$_{2}$ Exoplanetary Atmosphere Around an M-Dwarf Star

The chemical evolution of an exoplanetary Venus-like atmosphere is dependent upon the ultraviolet to near ultraviolet (FUV-NUV) radiation ratio from the parent star, the balance between CO$_{2}$ photolysis and recombination via reactions that depend on the water abundance, and various catalytic chemical cycles. In this study, we use a three-dimensional (3-D) model to simulate conditions for a Venus-like exoplanet orbiting the M-dwarf type star GJ 436 by varying the star/planet distance and considering the resultant effects on heating/cooling and dynamics. The simulation includes the middle and upper atmosphere (<40 mbar). Overall, these model comparisons reveal that the impact of extreme ultraviolet to ultraviolet (EUV-UV) heating on the energy balance shows both radiative and dynamical processes are responsible for driving significant variations in zonal winds and global temperature profiles at < 10$^{-5}$ mbar. More specifically, CO$_{2}$ 15-${\mu}$m cooling balances EUV/UV and Near InfraRed (NIR) heating at altitudes below 10$^{-7}$ mbar pressure with a strong maximum balance for pressures at ~10$^{-5}$ mbar, thus explaining the invariance of the temperature distribution at altitudes below 10$^{-5}$mbar pressure for all cases. Our model comparisons also show that moderate changes in NIR heating result in relatively small changes in neutral temperature in the upper atmosphere, and virtually no change in the middle atmosphere. However, with larger changes in the NIR heating profile, much greater changes in neutral temperature occur in the entire upper and middle atmosphere studied.

astro-ph.EP↗

arXiv2021

ST-PCNN: Spatio-Temporal Physics-Coupled Neural Networks for Dynamics Forecasting

Ocean current, fluid mechanics, and many other spatio-temporal physical dynamical systems are essential components of the universe. One key characteristic of such systems is that certain physics laws -- represented as ordinary/partial differential equations (ODEs/PDEs) -- largely dominate the whole process, irrespective of time or location. Physics-informed learning has recently emerged to learn physics for accurate prediction, but they often lack a mechanism to leverage localized spatial and temporal correlation or rely on hard-coded physics parameters. In this paper, we advocate a physics-coupled neural network model to learn parameters governing the physics of the system, and further couple the learned physics to assist the learning of recurring dynamics. A spatio-temporal physics-coupled neural network (ST-PCNN) model is proposed to achieve three goals: (1) learning the underlying physics parameters, (2) transition of local information between spatio-temporal regions, and (3) forecasting future values for the dynamical system. The physics-coupled learning ensures that the proposed model can be tremendously improved by using learned physics parameters, and can achieve good long-range forecasting (e.g., more than 30-steps). Experiments, using simulated and field-collected ocean current data, validate that ST-PCNN outperforms existing physics-informed models.

cs.LG↗

arXiv2018

Semisupervised Learning on Heterogeneous Graphs and its Applications to Facebook News Feed

Graph-based semi-supervised learning is a fundamental machine learning problem, and has been well studied. Most studies focus on homogeneous networks (e.g. citation network, friend network). In the present paper, we propose the Heterogeneous Embedding Label Propagation (HELP) algorithm, a graph-based semi-supervised deep learning algorithm, for graphs that are characterized by heterogeneous node types. Empirically, we demonstrate the effectiveness of this method in domain classification tasks with Facebook user-domain interaction graph, and compare the performance of the proposed HELP algorithm with the state of the art algorithms. We show that the HELP algorithm improves the predictive performance across multiple tasks, together with semantically meaningful embedding that are discriminative for downstream classification or regression tasks.

cs.SI↗

arXiv2015

Exponential Weight Functions for Quasi-Proportional Auctions

In quasi-proportional auctions, the allocation is shared among bidders in proportion to their weighted bids. The auctioneer selects a bid weight function, and bidders know the weight function when they bid. In this note, we analyze how weight functions that are exponential in the bid affect bidder behavior. We show that exponential weight functions have a pure-strategy Nash equilibrium, we characterize bids at an equilibrium, and we compare it to an equilibrium for power weight functions.

cs.GT↗

arXiv2015

Revenue-Maximizing Mechanism Design for Quasi-Proportional Auctions

In quasi-proportional auctions, each bidder receives a fraction of the allocation equal to the weight of their bid divided by the sum of weights of all bids, where each bid's weight is determined by a weight function. We study the relationship between the weight function, bidders' private values, number of bidders, and the seller's revenue in equilibrium. It has been shown that if one bidder has a much higher private value than the others, then a nearly flat weight function maximizes revenue. Essentially, threatening the bidder who has the highest valuation with having to share the allocation maximizes the revenue. We show that as bidder private values approach parity, steeper weight functions maximize revenue by making the quasi-proportional auction more like a winner-take-all auction. We also show that steeper weight functions maximize revenue as the number of bidders increases. For flatter weight functions, there is known to be a unique pure-strategy Nash equilibrium. We show that a pure-strategy Nash equilibrium also exists for steeper weight functions, and we give lower bounds for bids at an equilibrium. For a special case that includes the two-bidder auction, we show that the pure-strategy Nash equilibrium is unique, and we show how to compute the revenue at equilibrium. We also show that selecting a weight function based on private value ratios and number of bidders is necessary for a quasi-proportional auction to produce more revenue than a second-price auction.

cs.GT↗

arXiv2015

VCG Payments for Portfolio Allocations in Online Advertising

Some online advertising offers pay only when an ad elicits a response. Randomness and uncertainty about response rates make showing those ads a risky investment for online publishers. Like financial investors, publishers can use portfolio allocation over multiple advertising offers to pursue revenue while controlling risk. Allocations over multiple offers do not have a distinct winner and runner-up, so the usual second-price mechanism does not apply. This paper develops a pricing mechanism for portfolio allocations. The mechanism is efficient, truthful, and rewards offers that reduce risk.

cs.GT↗

arXiv2014

Validation of Matching

We introduce a technique to compute probably approximately correct (PAC) bounds on precision and recall for matching algorithms. The bounds require some verified matches, but those matches may be used to develop the algorithms. The bounds can be applied to network reconciliation or entity resolution algorithms, which identify nodes in different networks or values in a data set that correspond to the same entity. For network reconciliation, the bounds do not require knowledge of the network generation process.

cs.LG↗