Ye Ma

Zhihong Liu, Siqi Kou, Zheng Li et al.

Crafting a product display webpage from a source product image, along with layout and visual content instructions, holds significant practical value for domains such as marketing, advertising, and E-commerce. Intuitively, this task demands strict visual consistency across product displays and high-fidelity instruction following to jointly generate renderable HTML code. These requirements on controllability and instruction-following are closely aligned with the core features of advanced multimodal generative models, such as image editing models and unified models. To this end, this paper introduces ProductWebGen to systematically benchmark the product webpage generation capacities of these models. We organize ProductWebGen with 500 test samples covering 13 product categories; each sample consists of a source image, a visual content instruction, and a webpage instruction. The task is to generate a product showcase webpage including multiple consistent images in accordance with the source image and instructions. Given the mixed-modality input-output nature of the task, we design and systematically compare two workflows for evaluation -- one uses large language models and image editing models to separately generate HTML code and images (editing-based), while the other relies on a single UM to generate both, with image generation conditioned on the preceding multimodal context (UM-based). Empirical results show that editing-based approaches achieve leading results in webpage instruction following and content appeal, while UM-based ones may display more advantages in fulfilling visual content instructions. We also construct a supervised fine-tuning dataset, ProductWebGen-1k, with 1,000 groups of real product images and LLM-generated HTML code. We verify its effectiveness on the open-source UM BAGEL. The data and code are available at https://github.com/SJTU-DENG-Lab/ProductWebGen.

Jinpeng Lin, Min Zhou, Ye Ma et al. · amazon-science

Advertising posters, a form of information presentation, combine visual and linguistic modalities. Creating a poster involves multiple steps and necessitates design experience and creativity. This paper introduces AutoPoster, a highly automatic and content-aware system for generating advertising posters. With only product images and titles as inputs, AutoPoster can automatically produce posters of varying sizes through four key stages: image cleaning and retargeting, layout generation, tagline generation, and style attribute prediction. To ensure visual harmony of posters, two content-aware models are incorporated for layout and tagline generation. Moreover, we propose a novel multi-task Style Attribute Predictor (SAP) to jointly predict visual style attributes. Meanwhile, to our knowledge, we propose the first poster generation dataset that includes visual attribute annotations for over 76k posters. Qualitative and quantitative outcomes from user studies and experiments substantiate the efficacy of our system and the aesthetic superiority of the generated posters compared to other poster generation methods.

Yunning Cao, Ye Ma, Min Zhou et al.

Layout generation is a novel task in computer vision, which combines the challenges in both object localization and aesthetic appraisal, widely used in advertisements, posters, and slides design. An accurate and pleasant layout should consider both the intra-domain relationship within layout elements and the inter-domain relationship between layout elements and the image. However, most previous methods simply focus on image-content-agnostic layout generation, without leveraging the complex visual information from the image. To this end, we explore a novel paradigm entitled image-conditioned layout generation, which aims to add text overlays to an image in a semantically coherent manner. Specifically, we propose an Image-Conditioned Variational Transformer (ICVT) that autoregressively generates various layouts in an image. First, self-attention mechanism is adopted to model the contextual relationship within layout elements, while cross-attention mechanism is used to fuse the visual information of conditional images. Subsequently, we take them as building blocks of conditional variational autoencoder (CVAE), which demonstrates appealing diversity. Second, in order to alleviate the gap between layout elements domain and visual domain, we design a Geometry Alignment module, in which the geometric information of the image is aligned with the layout representation. In addition, we construct a large-scale advertisement poster layout designing dataset with delicate layout and saliency map annotations. Experimental results show that our model can adaptively generate layouts in the non-intrusive area of the image, resulting in a harmonious layout design.

Min Zhou, Chenchen Xu, Ye Ma et al.

In this paper, we study the graphic layout generation problem of producing high-quality visual-textual presentation designs for given images. We note that image compositions, which contain not only global semantics but also spatial information, would largely affect layout results. Hence, we propose a deep generative model, dubbed as composition-aware graphic layout GAN (CGL-GAN), to synthesize layouts based on the global and spatial visual contents of input images. To obtain training images from images that already contain manually designed graphic layout data, previous work suggests masking design elements (e.g., texts and embellishments) as model inputs, which inevitably leaves hint of the ground truth. We study the misalignment between the training inputs (with hint masks) and test inputs (without masks), and design a novel domain alignment module (DAM) to narrow this gap. For training, we built a large-scale layout dataset which consists of 60,548 advertising posters with annotated layout information. To evaluate the generated layouts, we propose three novel metrics according to aesthetic intuitions. Through both quantitative and qualitative evaluations, we demonstrate that the proposed model can synthesize high-quality graphic layouts according to image compositions.

Ye Ma, Mingming Sun, Ping Li

Recent years have witnessed the impressive progress in Neural Dependency Parsing. According to the different factorization approaches to the graph joint probabilities, existing parsers can be roughly divided into autoregressive and non-autoregressive patterns. The former means that the graph should be factorized into multiple sequentially dependent components, then it can be built up component by component. And the latter assumes these components to be independent so that they can be outputted in a one-shot manner. However, when treating the directed edge as an explicit dependency relationship, we discover that there is a mixture of independent and interdependent components in the dependency graph, signifying that both aforementioned models fail to precisely capture the explicit dependencies among nodes and edges. Based on this property, we design a Semi-Autoregressive Dependency Parser to generate dependency graphs via adding node groups and edge groups autoregressively while pouring out all group elements in parallel. The model gains a trade-off between non-autoregression and autoregression, which respectively suffer from the lack of target inter-dependencies and the uncertainty of graph generation orders. The experiments show the proposed parser outperforms strong baselines on Enhanced Universal Dependencies of multiple languages, especially achieving $4\%$ average promotion at graph-level accuracy. Also, the performances of model variations show the importance of specific parts.

Siqi Kou, Jiachun Jin, Zetong Zhou et al.

Recent progress in text-to-image (T2I) diffusion models (DMs) has enabled high-quality visual synthesis from diverse textual prompts. Yet, most existing T2I DMs, even those equipped with large language model (LLM)-based text encoders, remain text-pixel mappers -- they employ LLMs merely as text encoders, without leveraging their inherent reasoning capabilities to infer what should be visually depicted given the textual prompt. To move beyond such literal generation, we propose the think-then-generate (T2G) paradigm, where the LLM-based text encoder is encouraged to reason about and rewrite raw user prompts; the states of the rewritten prompts then serve as diffusion conditioning. To achieve this, we first activate the think-then-rewrite pattern of the LLM encoder with a lightweight supervised fine-tuning process. Subsequently, the LLM encoder and diffusion backbone are co-optimized to ensure faithful reasoning about the context and accurate rendering of the semantics via Dual-GRPO. In particular, the text encoder is reinforced using image-grounded rewards to infer and recall world knowledge, while the diffusion backbone is pushed to produce semantically consistent and visually coherent images. Experiments show substantial improvements in factual consistency, semantic alignment, and visual realism across reasoning-based image generation and editing benchmarks, achieving 0.79 on WISE score, nearly on par with GPT-4. Our results constitute a promising step toward next-generation unified models with reasoning, expression, and demonstration capacities.

Zixun Lan, Binjie Hong, Ye Ma et al.

Graph similarity measurement, which computes the distance/similarity between two graphs, arises in various graph-related tasks. Recent learning-based methods lack interpretability, as they directly transform interaction information between two graphs into one hidden vector and then map it to similarity. To cope with this problem, this study proposes a more interpretable end-to-end paradigm for graph similarity learning, named Similarity Computation via Maximum Common Subgraph Inference (INFMCS). Our critical insight into INFMCS is the strong correlation between similarity score and Maximum Common Subgraph (MCS). We implicitly infer MCS to obtain the normalized MCS size, with the supervision information being only the similarity score during training. To capture more global information, we also stack some vanilla transformer encoder layers with graph convolution layers and propose a novel permutation-invariant node Positional Encoding. The entire model is quite simple yet effective. Comprehensive experiments demonstrate that INFMCS consistently outperforms state-of-the-art baselines for graph-graph classification and regression tasks. Ablation experiments verify the effectiveness of the proposed computation paradigm and other components. Also, visualization and statistics of results reveal the interpretability of INFMCS.

Jingyu Liu, Minquan Wang, Ye Ma et al.

Videos showcasing specific products are increasingly important for E-commerce. Key moments naturally exist as the first appearance of a specific product, presentation of its distinctive features, the presence of a buying link, etc. Adding proper sound effects (SFX) to these key moments, or video decoration with SFX (VDSFX), is crucial for enhancing the user engaging experience. Previous studies about adding SFX to videos perform video to SFX matching at a holistic level, lacking the ability of adding SFX to a specific moment. Meanwhile, previous studies on video highlight detection or video moment retrieval consider only moment localization, leaving moment to SFX matching untouched. By contrast, we propose in this paper D&M, a unified method that accomplishes key moment detection and moment to SFX matching simultaneously. Moreover, for the new VDSFX task we build a large-scale dataset SFX-Moment from an E-commerce platform. For a fair comparison, we build competitive baselines by extending a number of current video moment detection methods to the new task. Extensive experiments on SFX-Moment show the superior performance of the proposed method over the baselines.

Ye Ma, Lu Zong

In comparison to single-document summarization, abstractive Multi-Document Summarization (MDS) brings challenges on the representation and coverage of its lengthy and linked sources. This study develops a Parallel Hierarchical Transformer (PHT) with attention alignment for MDS. By incorporating word- and paragraph-level multi-head attentions, the hierarchical architecture of PHT allows better processing of dependencies at both token and document levels. To guide the decoding towards a better coverage of the source documents, the attention-alignment mechanism is then introduced to calibrate beam search with predicted optimal attention distributions. Based on the WikiSum data, a comprehensive evaluation is conducted to test improvements on MDS by the proposed architecture. By better handling the inner- and cross-document information, results in both ROUGE and human evaluation suggest that our hierarchical model generates summaries of higher quality relative to other Transformer-based baselines at relatively low computational cost.

Jiangming Shu, Yuxiang Zhang, Ye Ma et al.

Retrieval-augmented agents can query external evidence, yet their reliability in multi-step reasoning remains limited: noisy retrieval may derail multi-hop question answering, while outcome-only reinforcement learning provides credit signals that are too coarse to optimize intermediate steps. We propose \textsc{EvalAct} (Evaluate-as-Action), which converts implicit retrieval quality assessment into an explicit action and enforces a coupled Search-to-Evaluate protocol so that each retrieval is immediately followed by a structured evaluation score, yielding process signals aligned with the interaction trajectory. To leverage these signals, we introduce Process-Calibrated Advantage Rescaling (PCAR), a GRPO-based optimization method that rescales advantages at the segment level according to evaluation scores, emphasizing reliable segments while updating uncertain ones conservatively. Experiments on seven open-domain QA benchmarks show that \textsc{EvalAct} achieves the best average accuracy, with the largest gains on multi-hop tasks, and ablations verify that the explicit evaluation loop drives the primary improvements while PCAR provides consistent additional benefits.

Siqi Kou, Jiachun Jin, Zhihong Liu et al.

We introduce Orthus, an autoregressive (AR) transformer that excels in generating images given textual prompts, answering questions based on visual inputs, and even crafting lengthy image-text interleaved contents. Unlike prior arts on unified multimodal modeling, Orthus simultaneously copes with discrete text tokens and continuous image features under the AR modeling principle. The continuous treatment of visual signals minimizes the information loss for both image understanding and generation while the fully AR formulation renders the characterization of the correlation between modalities straightforward. The key mechanism enabling Orthus to leverage these advantages lies in its modality-specific heads -- one regular language modeling (LM) head predicts discrete text tokens and one diffusion head generates continuous image features conditioning on the output of the backbone. We devise an efficient strategy for building Orthus -- by substituting the Vector Quantization (VQ) operation in the existing unified AR model with a soft alternative, introducing a diffusion head, and tuning the added modules to reconstruct images, we can create an Orthus-base model effortlessly (e.g., within mere 72 A100 GPU hours). Orthus-base can further embrace post-training to better model interleaved images and texts. Empirically, Orthus surpasses competing baselines including Show-o and Chameleon across standard benchmarks, achieving a GenEval score of 0.58 and an MME-P score of 1265.8 using 7B parameters. Orthus also shows exceptional mixed-modality generation capabilities, reflecting the potential for handling intricate practical generation tasks.

Shengyuan Liu, Bo Wang, Ye Ma et al.

Existing subject-driven text-to-image generation models suffer from tedious fine-tuning steps and struggle to maintain both text-image alignment and subject fidelity. For generating compositional subjects, it often encounters problems such as object missing and attribute mixing, where some subjects in the input prompt are not generated or their attributes are incorrectly combined. To address these limitations, we propose a subject-driven generation framework and introduce training-free guidance to intervene in the generative process during inference time. This approach strengthens the attention map, allowing for precise attribute binding and feature injection for each subject. Notably, our method exhibits exceptional zero-shot generation ability, especially in the challenging task of compositional generation. Furthermore, we propose a novel metric GroundingScore to evaluate subject alignment thoroughly. The obtained quantitative results serve as compelling evidence showcasing the effectiveness of our proposed method. The code will be released soon.

Zhihui Yin, Ye Ma, Xipeng Cao et al.

The proliferation of online short video platforms has driven a surge in user demand for short video editing. However, manually selecting, cropping, and assembling raw footage into a coherent, high-quality video remains laborious and time-consuming. To accelerate this process, we focus on a user-friendly new task called Video Moment Montage (VMM), which aims to accurately locate the corresponding video segments based on a pre-provided narration text and then arrange these video clips to create a complete video that aligns with the corresponding descriptions. The challenge lies in extracting precise temporal segments while ensuring intra-sentence and inter-sentence context consistency, as a single script sentence may require trimming and assembling multiple video clips. To address this problem, we present a novel \textit{Text-Video Multi-Grained Integration} method (TV-MGI) that efficiently fuses text features from the script with both shot-level and frame-level video features, which enables the global and fine-grained alignment between the video content and the corresponding textual descriptions in the script. To facilitate further research in this area, we introduce the Multiple Sentences with Shots Dataset (MSSD), a large-scale dataset designed explicitly for the VMM task. We conduct extensive experiments on the MSSD dataset to demonstrate the effectiveness of our framework compared to baseline methods.

Xueyuan Lin, Cehao Yang, Ye Ma et al.

Recently, large language models (LLMs) have demonstrated outstanding reasoning capabilities on mathematical and coding tasks. However, their application to financial tasks-especially the most fundamental task of stock movement prediction-remains underexplored. We study a three-class classification problem (up, hold, down) and, by analyzing existing reasoning responses, observe that: (1) LLMs follow analysts' opinions rather than exhibit a systematic, independent analytical logic (CoTs). (2) LLMs list summaries from different sources without weighing adversarial evidence, yet such counterevidence is crucial for reliable prediction. It shows that the model does not make good use of its reasoning ability to complete the task. To address this, we propose Reflective Evidence Tuning (RETuning), a cold-start method prior to reinforcement learning, to enhance prediction ability. While generating CoT, RETuning encourages dynamically constructing an analytical framework from diverse information sources, organizing and scoring evidence for price up or down based on that framework-rather than on contextual viewpoints-and finally reflecting to derive the prediction. This approach maximally aligns the model with its learned analytical framework, ensuring independent logical reasoning and reducing undue influence from context. We also build a large-scale dataset spanning all of 2024 for 5,123 A-share stocks, with long contexts (32K tokens) and over 200K samples. In addition to price and news, it incorporates analysts' opinions, quantitative reports, fundamental data, macroeconomic indicators, and similar stocks. Experiments show that RETuning successfully unlocks the model's reasoning ability in the financial domain. Inference-time scaling still works even after 6 months or on out-of-distribution stocks, since the models gain valuable insights about stock movement prediction.

Yuxiang Zhang, Jiangming Shu, Ye Ma et al.

Large Language Models face challenges in long-horizon agentic tasks as their constrained memory is easily overwhelmed by distracting or irrelevant context. Existing working memory methods typically rely on external, heuristic mechanisms that are decoupled from the agent's core policy. In this work, we reframe working memory management as a learnable, intrinsic capability. We propose a novel framework, Memory-as-Action, where an agent actively manages its working memory by executing explicit editing operations as part of a unified policy. This formulation allows an agent, trained via reinforcement learning, to balance memory curation against long-term task objectives under given resource constraints. However, such memory editing actions break the standard assumption of a continuously growing prefix in LLM interactions, leading to what we call trajectory fractures. These non-prefix changes disrupt the causal continuity required by standard policy gradient methods, making those methods inapplicable. To address this, we propose a new algorithm, Dynamic Context Policy Optimization, which enables stable end-to-end reinforcement learning by segmenting trajectories at memory action points and applying trajectory-level advantages to the resulting action segments. Our results demonstrate that jointly optimizing for task reasoning and memory management in an end-to-end fashion not only reduces overall computational consumption but also improves task performance, driven by adaptive context curation strategies tailored to the model's intrinsic capabilities.

Ye Ma, Jin Ma, Min Zhou et al.

The objective of image outpainting is to extend image current border and generate new regions based on known ones. Previous methods adopt generative adversarial networks (GANs) to synthesize realistic images. However, the lack of explicit semantic representation leads to blurry and abnormal image pixels when the outpainting areas are complex and with various objects. In this work, we decompose the outpainting task into two stages. Firstly, we train a GAN to extend regions in semantic segmentation domain instead of image domain. Secondly, another GAN model is trained to synthesize real images based on the extended semantic layouts. The first model focuses on low frequent context such as sizes, classes and other semantic cues while the second model focuses on high frequent context like color and texture. By this design, our approach can handle semantic clues more easily and hence works better in complex scenarios. We evaluate our framework on various datasets and make quantitative and qualitative analysis. Experiments demonstrate that our method generates reasonable extended semantic layouts and images, outperforming state-of-the-art models.

Ye Ma, Zixun Lan, Lu Zong et al.

This study develops a calibrated beam-based algorithm with awareness of the global attention distribution for neural abstractive summarization, aiming to improve the local optimality problem of the original beam search in a rigorous way. Specifically, a novel global protocol is proposed based on the attention distribution to stipulate how a global optimal hypothesis should attend to the source. A global scoring mechanism is then developed to regulate beam search to generate summaries in a near-global optimal fashion. This novel design enjoys a distinctive property, i.e., the global attention distribution could be predicted before inference, enabling step-wise improvements on the beam search through the global scoring mechanism. Extensive experiments on nine datasets show that the global (attention)-aware inference significantly improves state-of-the-art summarization models even using empirical hyper-parameters. The algorithm is also proven robust as it remains to generate meaningful texts with corrupted attention distributions. The codes and a comprehensive set of examples are available.

Ye Ma, Lu Zong, Peiwan Wang

In this study, a novel Distributed Representation of News (DRNews) model is developed and applied in deep learning-based stock market predictions. With the merit of integrating contextual information and cross-documental knowledge, the DRNews model creates news vectors that describe both the semantic information and potential linkages among news events through an attributed news network. Two stock market prediction tasks, namely the short-term stock movement prediction and stock crises early warning, are implemented in the framework of the attention-based Long Short Term-Memory (LSTM) network. It is suggested that DRNews substantially enhances the results of both tasks comparing with five baselines of news embedding models. Further, the attention mechanism suggests that short-term stock trend and stock market crises both receive influences from daily news with the former demonstrates more critical responses on the information related to the stock market {\em per se}, whilst the latter draws more concerns on the banking sector and economic policies.

Ye Ma, Lu Zong

Voluminous works have been implemented to exploit content-enhanced network embedding models, with little focus on the labelled information of nodes. Although TriDNR leverages node labels by treating them as node attributes, it fails to enrich unlabelled node vectors with the labelled information, which leads to the weaker classification result on the test set in comparison to existing unsupervised textual network embedding models. In this study, we design an integrated node encoder (INE) for textual networks which is jointly trained on the structure-based and label-based objectives. As a result, the node encoder preserves the integrated knowledge of not only the network text and structure, but also the labelled information. Furthermore, INE allows the creation of label-enhanced vectors for unlabelled nodes by entering their node contents. Our node embedding achieves state-of-the-art performances in the classification task on two public citation networks, namely Cora and DBLP, pushing benchmarks up by 10.0\% and 12.1\%, respectively, with the 70\% training ratio. Additionally, a feasible solution that generalizes our model from textual networks to a broader range of networks is proposed.

Yujie Qian, Yinpeng Dong, Ye Ma et al.

Measuring research impact and ranking academic achievement are important and challenging problems. Having an objective picture of research institution is particularly valuable for students, parents and funding agencies, and also attracts attention from government and industry. KDD Cup 2016 proposes the paper acceptance rank prediction task, in which the participants are asked to rank the importance of institutions based on predicting how many of their papers will be accepted at the 8 top conferences in computer science. In our work, we adopt a three-step feature engineering method, including basic features definition, finding similar conferences to enhance the feature set, and dimension reduction using PCA. We propose three ranking models and the ensemble methods for combining such models. Our experiment verifies the effectiveness of our approach. In KDD Cup 2016, we achieved the overall rank of the 2nd place.