arXiv subjects

Ze Zhao

Publications and source records attributed to Ze Zhao.

At least 19 recordsLinked to original sources

arXiv2026

TAPES: Task-Aligned Joint Predictive Pretraining for Transaction Event Sequences

Transaction event sequences support diverse prediction tasks in online commerce and digital financial services. The same transaction records can yield prediction targets with different semantics, including values, changes, and relations to past events. These targets impose different constraints on history representations, but their predictability alone does not establish the downstream utility of a target set. Moreover, target-wise prediction losses do not explicitly model conditional co-occurrence among targets. To address these challenges, we propose TAPES, a task-aligned joint predictive pretraining framework for transaction sequences. TAPES constructs a finite collection of candidate target sets from admissible field-operator combinations. It selects a target set according to the downstream utility of representations learned through short joint pretraining trials under matched budgets. It then reinitializes the encoder and pretrains it on the selected targets through a joint likelihood parameterized by a shared discrete latent variable, before downstream supervised adaptation. Experiments in a real-world transaction-risk setting and on multiple public financial benchmarks spanning diverse downstream tasks demonstrate that TAPES outperforms representative baselines. Ablation studies further validate the contributions of predictive target selection and joint modeling.

cs.LG↗

arXiv2026

InterFuse: Interaction-Guided Adaptive Fusion of Heterogeneous Models for Payment Risk Ranking

Payment platforms use view-specific models trained on transaction attributes, behavioral histories, and relational structures to rank transactions under limited review budgets. However, existing fusion methods often focus on prediction scores or interactions among model representations, with less attention to how these interactions should guide aggregation using each model's original prediction and input coverage. To address this limitation, we propose InterFuse, a fusion framework that uses a shared fusion state from cross-model interaction to guide transaction-adaptive aggregation. InterFuse encodes each model's representation, prediction score, and input coverage as a model-specific token, using learned states for unavailable outputs. Availability-aware attention updates these tokens and produces a shared fusion state. This state then guides aggregation over available models using their updated representations, original predictions, and input coverage. We evaluate InterFuse on an industrial payment replay and three public financial benchmarks using the same upstream models across fusion methods. InterFuse achieves the highest top-ranked recall and AUC on all four datasets, improving Recall@1% by 1.11-8.09 percentage points over the strongest baselines. Ablations show the benefits of cross-model interaction and adaptive aggregation, while industrial analyses find larger interaction gains when model predictions disagree. Additional analyses examine robustness to missing outputs and the trade-off between ranking gains and fusion latency.

cs.AI↗

arXiv2026

ID2Pattern: Pretraining Generalizable Identity Behavior Patterns for Transaction Risk Foundation Models

Large-scale behavioral pretraining offers a promising route to generalizable representations in online payment systems. However, high-cardinality identity fields such as card, device, and account IDs, though rich in behavioral signals, remain underexploited. Their continually growing vocabularies and limited semantic meaning make existing vocabulary-based and semantics-based encodings ill-suited. Their generalizable signal lies not in specific values, but in the identity behavior patterns they exhibit across a transaction sequence. Building on this insight, we propose ID2Pattern, a pretraining framework that brings these patterns into transaction foundation models. ID2Pattern first encodes identity fields with a Local Anonymous ID (LAID) tokenizer, which captures how each identity appears and recurs within the sequence, independent of its specific value. It then pretrains a multi-scale pattern codebook on unlabeled sequences, which abstracts LAID sequences into discrete behavior patterns for downstream prediction. Experiments on an industrial dataset with over 100M sequences and six public datasets show that ID2Pattern consistently outperforms strong baselines, improving industrial by 4.13% and public datasets by 8.83% on average. Notably, ID2Pattern remains equally effective even when all identity values are unseen during training, highlighting its potential to generalize to new users and markets.

cs.AI↗

arXiv2026

TRIDENT-Seq: Profile-Matched Prefix-Future Contrast for Label-Efficient Payment Risk Modeling

Ant International accumulates large volumes of transaction history across users and payment businesses. Confirmed fraud and chargeback outcomes, by contrast, are scarce, delayed, and specific to each business. A single encoder pretrained on unlabeled histories could serve all businesses, but only if pretraining captures the behavior that separates risk. Coarse activity differences can make contrastive pretraining on transaction logs too easy: transaction volume, amount, and timing can distinguish a history's continuation from arbitrary candidates without modeling how payment resources are used. We present TRIDENT-Seq, a transaction foundation model trained with Prefix-Future Contrast (PFC). PFC matches each prefix to a summary of its subsequent behavior among candidates with similar activity profiles; activity prediction and dispersion regularization act on the encoder output. Across three production payment businesses, the pretrained encoder improves average precision (AP) and recall at a 0.5% intervention budget over architecture-matched scratch training in all six business-budget settings. With a fixed validation set, it exceeds full-label scratch training in mean AP and recall using 10% of the training labels on Wallet and 25% on Shopping and SME. Pretraining only on another business's histories still helps. Replacing profile-matched candidates with random ones cuts recall from 24.62% to 5.64%. In Ant International's deployment evaluation, recorded amount-based chargeback and overall loss rates are 37.7% and 37.9% lower than those of the previous XGBoost-based system under the same intervention budget.

cs.LG↗

arXiv2026

Inductive Reasoning for Temporal Knowledge Graphs with Emerging Entities

Reasoning on Temporal Knowledge Graphs (TKGs) is essential for predicting future events and time-aware facts. While existing methods are effective at capturing relational dynamics, their performance is limited by a closed-world assumption, which fails to account for emerging entities not present in the training. Notably, these entities continuously join the network without historical interactions. Empirical study reveals that emerging entities are widespread in TKGs, comprising roughly 25\% of all entities. The absence of historical interactions of these entities leads to significant performance degradation in reasoning tasks. Whereas, we observe that entities with semantic similarities often exhibit comparable interaction histories, suggesting the presence of transferable temporal patterns. Inspired by this insight, we propose TransFIR (Transferable Inductive Reasoning), a novel framework that leverages historical interaction sequences from semantically similar known entities to support inductive reasoning. Specifically, we propose a codebook-based classifier that categorizes emerging entities into latent semantic clusters, allowing them to adopt reasoning patterns from similar entities. Experimental results demonstrate that TransFIR outperforms all baselines in reasoning on emerging entities, achieving an average improvement of 28.6% in Mean Reciprocal Rank (MRR) across multiple datasets. The implementations are available at https://github.com/zhaodazhuang2333/TransFIR.

cs.AI↗

arXiv2026

Efficient Document Parsing via Parallel Token Prediction

Document parsing, as a fundamental yet crucial vision task, is being revolutionized by vision-language models (VLMs). However, the autoregressive (AR) decoding inherent to VLMs creates a significant bottleneck, severely limiting parsing speed. In this paper, we propose Parallel-Token Prediction (PTP), a plugable, model-agnostic and simple-yet-effective method that enables VLMs to generate multiple future tokens in parallel with improved sample efficiency. Specifically, we insert some learnable tokens into the input sequence and design corresponding training objectives to equip the model with parallel decoding capabilities for document parsing. Furthermore, to support effective training, we develop a comprehensive data generation pipeline that efficiently produces large-scale, high-quality document parsing training data for VLMs. Extensive experiments on OmniDocBench and olmOCR-bench demonstrate that our method not only significantly improves decoding speed (1.6x-2.2x) but also reduces model hallucinations and exhibits strong generalization abilities.

cs.CL↗

arXiv2025

CHAINSFORMER: Numerical Reasoning on Knowledge Graphs from a Chain Perspective

Reasoning over Knowledge Graphs (KGs) plays a pivotal role in knowledge graph completion or question answering systems, providing richer and more accurate triples and attributes. As numerical attributes become increasingly essential in characterizing entities and relations in KGs, the ability to reason over these attributes has gained significant importance. Existing graph-based methods such as Graph Neural Networks (GNNs) and Knowledge Graph Embeddings (KGEs), primarily focus on aggregating homogeneous local neighbors and implicitly embedding diverse triples. However, these approaches often fail to fully leverage the potential of logical paths within the graph, limiting their effectiveness in exploiting the reasoning process. To address these limitations, we propose ChainsFormer, a novel chain-based framework designed to support numerical reasoning. Chainsformer not only explicitly constructs logical chains but also expands the reasoning depth to multiple hops. Specially, we introduces Relation-Attribute Chains (RA-Chains), a specialized logic chain, to model sequential reasoning patterns. ChainsFormer captures the step-by-step nature of multi-hop reasoning along RA-Chains by employing sequential in-context learning. To mitigate the impact of noisy chains, we propose a hyperbolic affinity scoring mechanism that selects relevant logic chains in a variable-resolution space. Furthermore, ChainsFormer incorporates an attention-based numerical reasoner to identify critical reasoning paths, enhancing both reasoning accuracy and transparency. Experimental results demonstrate that ChainsFormer significantly outperforms state-of-the-art methods, achieving up to a 20.0% improvement in performance. The implementations are available at https://github.com/zhaodazhuang2333/ChainsFormer.

cs.AI↗

arXiv2025

KPIs 2024 Challenge: Advancing Glomerular Segmentation from Patch- to Slide-Level

Chronic kidney disease (CKD) is a major global health issue, affecting over 10% of the population and causing significant mortality. While kidney biopsy remains the gold standard for CKD diagnosis and treatment, the lack of comprehensive benchmarks for kidney pathology segmentation hinders progress in the field. To address this, we organized the Kidney Pathology Image Segmentation (KPIs) Challenge, introducing a dataset that incorporates preclinical rodent models of CKD with over 10,000 annotated glomeruli from 60+ Periodic Acid Schiff (PAS)-stained whole slide images. The challenge includes two tasks, patch-level segmentation and whole slide image segmentation and detection, evaluated using the Dice Similarity Coefficient (DSC) and F1-score. By encouraging innovative segmentation methods that adapt to diverse CKD models and tissue conditions, the KPIs Challenge aims to advance kidney pathology analysis, establish new benchmarks, and enable precise, large-scale quantification for disease research and diagnosis.

cs.CV↗

arXiv2024

Hunyuan-Large: An Open-Source MoE Model with 52 Billion Activated Parameters by Tencent

In this paper, we introduce Hunyuan-Large, which is currently the largest open-source Transformer-based mixture of experts model, with a total of 389 billion parameters and 52 billion activation parameters, capable of handling up to 256K tokens. We conduct a thorough evaluation of Hunyuan-Large's superior performance across various benchmarks including language understanding and generation, logical reasoning, mathematical problem-solving, coding, long-context, and aggregated tasks, where it outperforms LLama3.1-70B and exhibits comparable performance when compared to the significantly larger LLama3.1-405B model. Key practice of Hunyuan-Large include large-scale synthetic data that is orders larger than in previous literature, a mixed expert routing strategy, a key-value cache compression technique, and an expert-specific learning rate strategy. Additionally, we also investigate the scaling laws and learning rate schedule of mixture of experts models, providing valuable insights and guidances for future model development and optimization. The code and checkpoints of Hunyuan-Large are released to facilitate future innovations and applications. Codes: https://github.com/Tencent/Hunyuan-Large Models: https://huggingface.co/tencent/Tencent-Hunyuan-Large

cs.CL↗

arXiv2024

Cross-Organ and Cross-Scanner Adenocarcinoma Segmentation using Rein to Fine-tune Vision Foundation Models

In recent years, significant progress has been made in tumor segmentation within the field of digital pathology. However, variations in organs, tissue preparation methods, and image acquisition processes can lead to domain discrepancies among digital pathology images. To address this problem, in this paper, we use Rein, a fine-tuning method, to parametrically and efficiently fine-tune various vision foundation models (VFMs) for MICCAI 2024 Cross-Organ and Cross-Scanner Adenocarcinoma Segmentation (COSAS2024). The core of Rein consists of a set of learnable tokens, which are directly linked to instances, improving functionality at the instance level in each layer. In the data environment of the COSAS2024 Challenge, extensive experiments demonstrate that Rein fine-tuned the VFMs to achieve satisfactory results. Specifically, we used Rein to fine-tune ConvNeXt and DINOv2. Our team used the former to achieve scores of 0.7719 and 0.7557 on the preliminary test phase and final test phase in task1, respectively, while the latter achieved scores of 0.8848 and 0.8192 on the preliminary test phase and final test phase in task2. Code is available at GitHub.

eess.IV↗

arXiv2024

OXYGENERATOR: Reconstructing Global Ocean Deoxygenation Over a Century with Deep Learning

Accurately reconstructing the global ocean deoxygenation over a century is crucial for assessing and protecting marine ecosystem. Existing expert-dominated numerical simulations fail to catch up with the dynamic variation caused by global warming and human activities. Besides, due to the high-cost data collection, the historical observations are severely sparse, leading to big challenge for precise reconstruction. In this work, we propose OxyGenerator, the first deep learning based model, to reconstruct the global ocean deoxygenation from 1920 to 2023. Specifically, to address the heterogeneity across large temporal and spatial scales, we propose zoning-varying graph message-passing to capture the complex oceanographic correlations between missing values and sparse observations. Additionally, to further calibrate the uncertainty, we incorporate inductive bias from dissolved oxygen (DO) variations and chemical effects. Compared with in-situ DO observations, OxyGenerator significantly outperforms CMIP6 numerical simulations, reducing MAPE by 38.77%, demonstrating a promising potential to understand the "breathless ocean" in data-driven manner.

cs.LG↗

arXiv2023

Graph Out-of-Distribution Generalization with Controllable Data Augmentation

Graph Neural Network (GNN) has demonstrated extraordinary performance in classifying graph properties. However, due to the selection bias of training and testing data (e.g., training on small graphs and testing on large graphs, or training on dense graphs and testing on sparse graphs), distribution deviation is widespread. More importantly, we often observe \emph{hybrid structure distribution shift} of both scale and density, despite of one-sided biased data partition. The spurious correlations over hybrid distribution deviation degrade the performance of previous GNN methods and show large instability among different datasets. To alleviate this problem, we propose \texttt{OOD-GMixup} to jointly manipulate the training distribution with \emph{controllable data augmentation} in metric space. Specifically, we first extract the graph rationales to eliminate the spurious correlations due to irrelevant information. Secondly, we generate virtual samples with perturbation on graph rationale representation domain to obtain potential OOD training samples. Finally, we propose OOD calibration to measure the distribution deviation of virtual samples by leveraging Extreme Value Theory, and further actively control the training distribution by emphasizing the impact of virtual OOD samples. Extensive studies on several real-world datasets on graph classification demonstrate the superiority of our proposed method over state-of-the-art baselines.

cs.LG↗

arXiv2023

Drag on a partially immersed sphere at the capillary scale

We study the drag on a centimetric sphere in a uniform flow in the presence of a free surface as a function of submergence depth. Through direct force measurements in a custom benchtop recirculating flume, we demonstrate that the drag can significantly exceed the corresponding drag in a single-phase flow and achieves a peak at submergence depths just prior to complete immersion. The additional drag in the partially immersed state is rationalized by considering hydrostatic effects associated with the asymmetric surface height profile induced by the obstacle in the flow direction which persists for flow speeds below the minimum capillary-gravity wave speed. At these scales, the sphere's wettability plays a pronounced role in determining the maximum possible drag and results in hysteretic behaviors near touchdown and complete immersion. The influence of flow speed, sphere size, and surface tension on the drag characteristics are additionally explored through a combination of experiments and numerical simulations.

physics.flu-dyn↗

arXiv2022

Computational thermal multi-phase flow for metal additive manufacturing

Thermal multi-phase flow simulations are indispensable to understanding the multi-scale and multi-physics phenomena in metal additive manufacturing (AM) processes, yet accurate and robust predictions remain challenging. This book chapter summarizes the recent method development at UIUC for simulating thermal multiphase flows in laser powder bed fusion (LPBF) and directed energy deposition (DED) processes. Two main method developments are discussed. The first is a mixed interface-capturing/interface-tracking computational framework aiming to explicitly treat the gas-metal interface without mesh motion/re-meshing. The second is a physics-based and non-empirical deposit geometry model for DED processes. The proposed framework's accuracy is assessed by thoroughly comparing the simulated results against experimental measurements on various quantities. We also report critical quantities that experiments can not measure to show the predictive capability of the developed methods.

cs.CE↗

arXiv2020

Dynamical mixing between $2^3S_1$ and $1^3D_1$ charmed mesons

In charmed $D$ and $D_s$ mesons sector, the matrix of a Hamiltonian in a quark potential model is computed in the $2^3S_1$ and $1^3D_1$ subspace. The masses of four mixed states of $2^3S_1$ and $1^3D_1$ denoted with $D^*_1(2635)$, $D^*_1(2739)$, $D^*_{s1}(2715)$ and $D^*_{s1}(2805)$ are obtained. It is an off-diagonal part of the spin-orbit tensor interaction that causes the mixing between the $2^3S_1$ and $1^3D_1$ states. The mixing angles between the $2^3S_1$ and $1^3D_1$ states are tiny. Under the mixing, a $^3P_0$ model is employed to compute the hadronic decay widths of all OZI-allowed decay channels of the four mixed states. The two light mixed states $D^*_1(2635)$ and $D^*_{s1}(2715)$ are close in mass to $D^*_J(2600)$ and $D^*_{s1}(2700)$, while the two heavy mixed states $D^*_1(2739)$ and $D^*_{s1}(2805)$ are lighter in mass than $D(2750)$ and $D^*_{s1}(2860)$. The mixing angles obtained from dynamical interaction are inconsistent with the mixing angles obtained from hadronic decay. Based on mass spectra and hadronic decay analyses, $D^*_J(2600)$, $D(2750)$, $D^*_{s1}(2700)$ and $D^*_{s1}(2860)$ are impossibly the mixed states of $2^3S_1$ and $1^3D_1$ at the small mixing angles. The inconsistence implies that $D^*_1(2760)$ and $D^*_{s1}(2860)$ have not been properly resolved from present experimental data, or there exist large unknown off-diagonal interactions that result in large mixing angles.

hep-ph↗

arXiv2020

Theoretical interpretation of $Ξ_c(2970)$

The open charm strong decay widths and certain ratio of branching fractions of a charmed strange baryon $Ξ_c(2970)$ are calculated in a $^3P_0$ model. The results are compatible with the latest experimental data. The theoretical ratio of decay branching fractions $R = \mathcal{B} [ Ξ_{c}(2970)^+ \rightarrow Ξ_c(2645)^{0}π^{+} ] / \mathcal{B} [ Ξ_{c}(2970)^+ \rightarrow Ξ_c^{\prime0}π^{+}] \approx 1.0$. The spin-parity $J^P = 1/2^+$ and $3/2^+$ for different assignments are analyzed. From the results of our calculation, $Ξ_{c}(2970)$ can be interpreted as a $2S$-wave state with $J^P(s_l) = 1/2^+(0)$. The distinguishing between the 2S-wave $n_ρ$- and $n_λ$-excitation states and between states with $s_l = 0$ and $s_l = 1$ and between states with total spin $1/2$ and $3/2$($s_l = 1$) are also discussed.

hep-ph↗

arXiv2017

Study of $P$-wave excitations of observed charmed strange baryons

Many excited charmed strange baryons such as $Ξ_c(2790)$, $Ξ_c(2815)$, $Ξ_c(2930)$, $Ξ_c(2980)$, $Ξ_c(3055)$, $Ξ_c(3080)$ and $Ξ_c(3123)$ have been observed. In order to understand their internal structure and to determine their spin-parities, the strong decay properties of these baryons as possible $P$-wave excited $Ξ_c$ candidates have been systematically studied in a $^3P_0$ model. The configurations and $J^P$ assignments of $Ξ_c(2790)$, $Ξ_c(2815)$, $Ξ_c(2930)$, $Ξ_c(2980)$, $Ξ_c(3055)$, $Ξ_c(3080)$ and $Ξ_c(3123)$ have been explored based on recent experimental data. In our analyses, $Ξ_c(3055)$, $Ξ_c(3080)$ and $Ξ_c(3123)$ seem impossible to be the $P$-wave excited $Ξ_c$. $Ξ_c(2790)$, $Ξ_c(2815)$, $Ξ_c(2930)$ and $Ξ_c(2980)$ may be the $P$-wave excited $Ξ_c$. In particular, $Ξ_c(2790)$ and $Ξ_c(2815)$ are very possibly the $P$-wave excited $Ξ_{c1}(1/2^-)$ and $Ξ_{c1}(3/2^-)$, respectively. $Ξ_c(2980)$ may be the $P$-wave excited $Ξ_{c1}^{'}(\frac{1}{2}^-)$. $Ξ_c(2930)$ may be the $P$-wave $Ξ_{c0}^{'}(\frac{1}{2}^-)$, $\tildeΞ_{c0}(\frac{1}{2}^-)$, $Ξ_{c2}^{'}(\frac{3}{2}^-)$, $Ξ_{c2}^{'}(\frac{5}{2}^-)$, $\tildeΞ_{c2}(\frac{3}{2}^-)$ or $\tildeΞ_{c2}(\frac{5}{2}^-)$. Furthermore, some branching fraction ratios related to the internal structure and quark configuration of $P$-wave $Ξ_c$ have also been computed. Measurements of these ratios in the future will be helpful to understand these excited $Ξ_c$.

hep-ph↗

arXiv2017

Study of $2S$- and $1D$- excitations of observed charmed strange baryons

Strong decays of $Ξ_c$ baryons with radial or orbital excitations with positive parity have been studied in a $^3P_0$ model. As candidates of these $Xi_c$, possible configurations and $J^P$ of $Ξ_c(2930)$, $Ξ_c(2980)$, $Ξ_c(3055)$, $Ξ_c(3080)$ and $Ξ_c(3123)$ have been assigned. There are $40$ kinds of configurations to describe these excited $Ξ_c$. In these assignments, $Ξ_c(2930)$ may be a $2S$-wave excited $\tildeΞ_{c1}(\frac{1}{2}^+)$ or $\tildeΞ_{c1}(\frac{3}{2}^+)$, or a $D$-wave excited $\hatΞ_{c1}^{' }(\frac{1}{2}^+)$, $\checkΞ_{c1}^{\ 0}(\frac{1}{2}^+)$, $\checkΞ_{c1}^{\ 2}(\frac{1}{2}^+)$, $\hatΞ_{c1}^{' }(\frac{3}{2}^+)$, $\checkΞ_{c1}^{\ 0}(\frac{3}{2}^+)$ or $\checkΞ_{c1}^{\ 2}(\frac{3}{2}^+)$. $Ξ_c(2980)^+$ may be a $2S$-wave excited $\tildeΞ_{c1}(\frac{1}{2}^+)$ or $\tildeΞ_{c0}^{'}(\frac{1}{2}^+)$ with $J^P={1\over 2}^+$, or a $D$-wave excited $\checkΞ_{c0}^{'0}(\frac{1}{2}^+)$ or $\checkΞ_{c1}^{\ 0}(\frac{1}{2}^+)$ with $J^P={1\over 2}^+$. $Ξ_c(3055)^+$ may be a $2S$-wave excited $\acuteΞ_{c1}^{'}(\frac{3}{2}^+)$ or $\acuteΞ_{c0}(\frac{1}{2}^+)$. It may be a $D$-wave excited $Ξ_{c1}^{' }(\frac{3}{2}^+)$, $Ξ_{c2}^{' }(\frac{5}{2}^+)$, $Ξ_{c2}^{ }(\frac{3}{2}^+)$ or $Ξ_{c2}^{ }(\frac{5}{2}^+)$. $Ξ_c(3080)^+$ is very possibly a $2S$-wave excited $\acuteΞ_{c0}(\frac{1}{2}^+)$, and seems not a $D$-wave excitation of $Ξ_c$. For the poor experimental information of $Ξ_c(3123)$, it is impossible to identify this state at present. It is found that the channel $ΛD$ vanishes in the strong decay of $P$-wave, $D$-wave and $2S$-wave excited $Ξ_c$ without $ρ$- mode excitation between the two light quarks ($n_ρ=L_ρ=0$). Some branching fraction ratios have been computed and can be employed to distinguish different configurations in forthcoming experiments.

hep-ph↗