In the AIGC era, generating high-quality 4D content has garnered increasing research attention. Unfortunately, current 4D synthesis research is severely constrained by the lack of large-scale 4D datasets, preventing models from adequately learning the critical spatial-temporal features necessary for high-quality 4D generation, thus hindering progress in this domain. To combat this, we propose a novel framework that transfers rich spatial priors from existing 3D diffusion models and temporal priors from video diffusion models to enhance 4D synthesis. We develop a spatial-temporal-disentangled 4D (STD-4D) Diffusion model, which synthesizes 4D-aware videos through disentangled spatial and temporal latents. To facilitate the best feature transfer, we design a novel Orthogonal Spatial-temporal Distributional Transfer (Orster) mechanism, where the spatiotemporal feature distributions are carefully modeled and injected into the STD-4D Diffusion. Further, during the 4D construction, we devise a spatial-temporal-aware HexPlane (ST-HexPlane) to integrate the transferred spatiotemporal features for better 4D deformation and 4D Gaussian feature modeling. Experiments demonstrate that our method significantly outperforms existing approaches, achieving superior spatial-temporal consistency and higher-quality 4D synthesis.

Bobo Li, Wu Rui, Zibo Ji, Meishan Zhang, Hao Fei, Min Zhang, Mong-Li Lee, Wynne Hsu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Automated fact-checking aims to verify the veracity of claims based on related evidence, and has become increasingly important as large language models (LLMs) make it easier to generate and disseminate misinformation at scale. In open settings, effective fact-checking requires models to iteratively retrieve relevant evidence and reason over noisy and incomplete information. While recent LLM-based approaches have shown promising reasoning capabilities, prompt-based methods remain limited by the inherent behaviors of base LLMs, and supervised fine-tuning methods typically require costly annotated reasoning trajectories. In this paper, we propose R3Check, a rule-guided reinforcement learning framework that enables LLMs to perform iterative retrieval–reasoning for multi-hop fact-checking. R3Check formulates the retriever as an external environment and optimizes the model using Group Relative Policy Optimization, relying only on final veracity labels and format-based rewards rather than explicit reasoning annotations. To mitigate the mutual interference between retrieval and reasoning that arises under joint training, we introduce a two-stage curriculum that first trains structured reasoning under closed fact-checking with gold evidence, and then jointly optimizes retrieval and reasoning with real-time retrieval. An importance-based sampling strategy further strengthens effective supervision signals during training. Despite using only a 7B backbone, R3Check outperforms existing baselines and even powerful reasoning LLMs, under both given-evidence and real-time retrieval settings, while producing interpretable reasoning chains. This work demonstrates the potential of pure reinforcement learning to induce effective retrieval–reasoning behaviors for fact-checking under weak supervision.

Vision-language models (VLMs) exhibit impressive zeroshot transfer, but remain static and cannot adapt when exposed to new tasks. Meanwhile, conventional continual learning often overlooks preserving this zero-shot capability during adaptation. In this work, we present DREAM, a parameter-efficient framework that enables continual adaptation of VLMs while minimizing forgetting and preserving zero-shot performance. DREAM employs a dynamic prompt system with lightweight, task-specific parameters managed by two modules: a prompt composition module that dynamically generates prompts to adapt the VLM, and a query-key module that uses learned token weights to reliably activate the appropriate parameters at inference. To enhance robustness, we propose GuidedMix, which creates semantically meaningful mixed images, and pair them with mixture-aware text embeddings to strengthen representation learning through image-text alignment. We further leverage the GuidedMix samples to estimate task-specific query-key similarity thresholds that identify samples of unseen tasks and and prevent spurious prompt usage on the VLM, thereby safeguarding its zero-shot behavior. Experiments show that our method adapts efficiently, mitigates forgetting, and maintains strong zero-shot transfer with substantially fewer trainable parameters, showing consistent gains even under partial supervision.

Jundong Xu, Hao Fei, Meng Luo, Qian Liu, Liangming Pan, William Yang Wang, Preslav Nakov, Mong-Li Lee, Wynne Hsu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Human Activity Recognition (HAR) is pivotal for behavior monitoring in public healthcare, supporting tasks like medical rehabilitation and targeted wellness campaigns. Initially utilizing hand-crafted features and traditional machine learning models such as support vector machines and decision trees, HAR has since evolved to include deep learning techniques, especially convolutional neural networks and Transformers, which excel at modeling temporal and task-specific information from sensor data. Despite these advancements, challenges related to high task dependency and data scarcity persist. To address these issues, there have been efforts to harness the vast pre-trained knowledge of Large Language Models (LLMs) for HAR. Yet, LLMs often fail to fully capture the temporal dynamics inherent in sensor data. We introduce HyMv, a novel hybrid approach that combines an auxiliary HAR model with soft-prompt tuning of LLMs. This approach leverages the auxiliary model’s proficiency in processing sensor data to guide the parameter optimization of LLMs during prompt tuning. Importantly, the auxiliary HAR model is only active during training, augmenting the LLM’s parameters without adding computational overhead during testing. We evaluate the performance of HyMv for various HAR tasks on diverse datasets and demonstrate its adaptability to different sensor modalities, further showcasing its broad applicability.

Continual learning is essential for adapting models to new tasks while retaining previously acquired knowledge. While existing approaches predominantly focus on uni-modal data, multi-modal learning offers substantial benefits by utilizing diverse sensory inputs, akin to human perception. However, multi-modal continual learning presents additional challenges, as the model must effectively integrate new information from various modalities while preventing catastrophic forgetting. In this work, we propose a pre-trained model-based framework for multi-modal continual learning. Our framework includes a novel cross-modality adapter with a mixture-of-experts structure to facilitate effective integration of multi-modal information across tasks. We also introduce a representation alignment loss that fosters learning of robust multi-modal representations, and regularize relationships between learned representations to preserve knowledge from previous tasks. Experiments on several multi-modal datasets demonstrate that our approach consistently outperforms baselines in both class-incremental and domain-incremental learning, achieving higher accuracy and reduced forgetting.

The rapid advancement of Large Language Models (LLMs) has revolutionized the generation of emotional support conversations (ESC), offering scalable solutions with reduced costs and enhanced data privacy.This paper explores the role of personas in the creation of ESC by LLMs.Our research utilizes established psychological frameworks to measure and infuse persona traits into LLMs, which then generate dialogues in the emotional support scenario.We conduct extensive evaluations to understand the stability of persona traits in dialogues, examining shifts in traits post-generation and their impact on dialogue quality and strategy distribution.Experimental results reveal several notable findings: 1) LLMs can infer core persona traits, 2) subtle shifts in emotionality and extraversion occur, influencing the dialogue dynamics, and 3) the application of persona traits modifies the distribution of emotional support strategies, enhancing the relevance and empathetic quality of the responses.These findings highlight the potential of persona-driven LLMs in crafting more personalized, empathetic, and effective emotional support dialogues, which has significant implications for the future design of AI-driven emotional support systems.

Multimodal misinformation, encompassing textual, visual, and cross-modal distortions, poses an increasing societal threat that is amplified by generative AI. Existing methods typically focus on a single type of distortion and struggle to generalize to unseen scenarios. In this work, we observe that different distortion types share common reasoning capabilities while also requiring task-specific skills. We hypothesize that joint training across distortion types facilitates knowledge sharing and enhances the model’s ability to generalize. To this end, we introduce TRUST-VL, a unified and explainable vision-language model for general multimodal misinformation detection. TRUST-VL incorporates a novel Question-Aware Visual Amplifier module, designed to extract task-specific visual features. To support training, we also construct TRUST-Instruct, a large-scale instruction dataset containing 198K samples featuring structured reasoning chains aligned with human fact-checking workflows. Extensive experiments on both in-domain and zero-shot benchmarks demonstrate that TRUST-VL achieves state-of-the-art performance, while also offering strong generalization and interpretability.

Multi-Modal Large Language Models (MLLMs) have exhibited remarkable performance on various vision-language tasks such as Visual Question Answering (VQA). Despite accumulating evidence of privacy concerns associated with task-relevant content, it remains unclear whether MLLMs inadvertently memorize private content that is entirely irrelevant to the training tasks. In this paper, we investigate how randomly generated task-irrelevant private content can become spuriously correlated with downstream objectives due to partial mini-batch training dynamics, thus causing inadvertent memorization. Concretely, we randomly generate task-irrelevant watermarks into VQA fine-tuning images at varying probabilities and propose a novel probing framework to determine whether MLLMs have inadvertently encoded such content. Our experiments reveal that MLLMs exhibit notably different training behaviors in partial mini-batch settings with task-irrelevant watermarks embedded. Furthermore, through layer-wise probing, we demonstrate that MLLMs trigger distinct representational patterns when encountering previously seen task-irrelevant knowledge, even if this knowledge does not influence their output during prompting. Our code is available at https://github.com/illusionhi/ProbingPrivacy.

Temporal claims, often riddled with inaccuracies, are a significant challenge in the digital misinformation landscape. Fact-checking systems that can accurately verify such claims are crucial for combating misinformation. Current systems struggle with the complexities of evaluating the accuracy of these claims, especially when they include multiple, overlapping, or recurring events. We introduce a novel timeline-based fact verification framework that identify events from both claim and evidence and organize them into their respective chronological timelines. The framework systematically examines the relationships between the events in both claim and evidence to predict the veracity of each claim event and their chronological accuracy. This allows us to accurately determine the overall veracity of the claim. We also introduce a new dataset of complex temporal claims involving timeline-based reasoning for the training and evaluation of our proposed framework. Experimental results demonstrate the effectiveness of our approach in handling the intricacies of temporal claim verification.

Recent progress of generative AI and the popularity of short-form video-sharing platforms have raised new risks of misinformation video issues, posing a potential threat to online multimedia ecosystems. With the aid of generative AI tools, producing and spreading vivid, persuasive misinformation videos has been easier, while detecting and preventing them has become harder. This tutorial introduces how to characterize, detect, and prevent misinformation videos, which consists of three technical parts: 1) Characterization of AI-generated and human-edited misinformation videos; 2) Detection approaches, covering those tailored for fully generated, manipulated, and human-edited videos; and 3) Prevention strategies, including those effective for the creation and spread phases. This tutorial concludes by discussing the status quo and ongoing challenges and highlighting the promising directions for future research. We expect to bring broader attention to misinformation video issues, gather and communicate with researchers of interest, and facilitate the engagement of those who are new to this field.

Online form filling is a common yet labor-intensive task involving extensive keyboard and mouse interactions. Despite the long-standing vision of automating this process with ''one click,'' existing tools remain largely rule-based and lack generalizable, generative capabilities. Recent advances in Multimodal Large Language Models (MLLMs) have enabled promising agents for GUI-related tasks in general-purpose scenarios. However, they struggle with the unique challenges of form filling, such as flexible layouts and the difficulty of aligning textual instructions with on-screen fields. To bridge this gap, we formally define the form-filling task and propose FormFactory-an interactive benchmarking suite comprising a web-based interface, backend evaluation module, and carefully constructed dataset. Our benchmark covers diverse real-world scenarios, incorporates various field formats, and simulates high-fidelity form interactions. We conduct a comprehensive evaluation of state-of-the-art MLLMs and observe that no model surpasses 5% accuracy, underscoring the inherent difficulty of the task. These findings also reveal significant limitations in current models' visual layout reasoning and field-value alignment abilities.

Understanding fine-grained sentiment dynamics in human conversations is a central goal for next-generation artificial intelligence, especially in scenarios where interactions are rich in both modalities and context. To advance research in this area, we organize the Multimodal Conversational Aspect-based Sentiment Analysis (MCABSA) challenge to the community of aspect-based sentiment analysis. The MCABSA challenge introduces two novel subtasks: 1) Panoptic Sentiment Sextuple Extraction, panoramically recognizing holder, target, aspect, opinion, sentiment, and rationale from multi-turn, multi-party multimodal dialogue; and 2) Sentiment Flipping Analysis, detecting the dynamic sentiment transformation throughout the conversation along with the causal reasons. To support these tasks, we present the PanoSent dataset, a high-quality, large-scale benchmark featuring multi-turn, multi-party dialogues annotated with both explicit and implicit sentiment elements across text, image, audio, and video modalities. PanoSent offers extensive real-world scenario coverage, providing a comprehensive testbed for multimodal conversational sentiment analysis. The challenge has attracted widespread participation from both academia and industry, with over 30 teams registered and more than 100 successful submissions. In this paper, we introduce the task, dataset, and evaluation settings, summarize the systems of the top teams, and discuss the findings of the participants. Further details of the challenge can be found at https://panosent.github.io/MM25-challenge.

RGB-D salient object detection (SOD) aims to identify the most conspicuous objects in a scene with the incorporation of depth cues. Existing methods mainly rely on CNNs, limited by the local receptive fields, or Vision Transformers that suffer from the cost of quadratic complexity, posing a challenge in balancing performance and computational efficiency. Recently, state space models (SSM), Mamba, have shown great potential for modeling long-range dependency with linear complexity. However, directly applying SSM to RGB-D SOD may lead to deficient local semantics as well as the inadequate cross-modality fusion. To address these issues, we propose a Local Emphatic and Adaptive Fusion state space model (LEAF-Mamba) that contains two novel components: 1) a local emphatic state space module (LE-SSM) to capture multi-scale local dependencies for both modalities. 2) an SSM-based adaptive fusion module (AFM) for complementary cross-modality interaction and reliable cross-modality integration. Extensive experiments demonstrate that the LEAF-Mamba consistently outperforms 16 state-of-the-art RGB-D SOD methods in both efficacy and efficiency. Moreover, our method can achieve excellent performance on the RGB-T SOD task, proving a powerful generalization ability. Our code is publicly available at https://github.com/LanhooNg/LEAF-Mamba.

Test-time adaptation aims to improve model robustness under distribution shifts by adapting models with access to unlabeled target samples. A primary cause of performance degradation under such shifts is the model’s reliance on features that lack a direct causal relationship with the prediction target. We introduce Test-time Adaptation by Causal Trimming (TACT), a method that identifies and removes non-causal components from representations for test distributions. TACT applies data augmentations that preserve causal features while varying non-causal ones. By analyzing the changes in the representations using Principal Component Analysis, TACT identifies the highest variance directions associated with non-causal features. It trims the representations by removing their projections on the identified directions, and uses the trimmed representations for the predictions. During adaptation, TACT continuously tracks and refines these directions to get a better estimate of non-causal features. We theoretically analyze the effectiveness of this approach and empirically validate TACT on real-world out-of-distribution benchmarks. TACT consistently outperforms state-of-the-art methods by a significant margin.

Multimodal symbolic logical reasoning, which aims to deduce new facts from multimodal input via formal logic, is critical in high-stakes applications such as autonomous driving and medical diagnosis, as its rigorous, deterministic reasoning helps prevent serious consequences. To evaluate such capabilities of current state-of-the-art vision language models (VLMs), we introduce the first benchmark MuSLR for multimodal symbolic logical reasoning grounded in formal logical rules. MuSLR comprises 1,093 instances across 7 domains, including 35 atomic symbolic logic and 976 logical combinations, with reasoning depths ranging from 2 to 9. We evaluate 7 state-of-the-art VLMs on MuSLR and find that they all struggle with multimodal symbolic reasoning, with the best model, GPT-4.1, achieving only 46.8%. Thus, we propose LogiCAM, a modular framework that applies formal logical rules to multimodal inputs, boosting GPT-4.1’s Chain-of-Thought performance by 14.13%, and delivering even larger gains on complex logics such as first-order logic. We also conduct a comprehensive error analysis, showing that around 70% of failures stem from logical misalignment between modalities, offering key insights to guide future improvements.

While the recent Chain-of-Thought (CoT) technique enhances the reasoning ability of large language models (LLMs) with the theory of mind, it might still struggle in handling logical reasoning that relies much on symbolic expressions and rigid deducing rules.To strengthen the logical reasoning capability of LLMs, we propose a novel Symbolic Chain-of-Thought, namely SymbCoT, a fully LLM-based framework that integrates symbolic expressions and logic rules with CoT prompting.Technically, building upon an LLM, SymbCoT 1) first translates the natural language context into the symbolic format, and then 2) derives a step-by-step plan to solve the problem with symbolic logical rules, 3) followed by a verifier to check the translation and reasoning chain.Via thorough evaluations on 5 standard datasets with both First-Order Logic and Constraint Optimization symbolic expressions, SymbCoT shows striking improvements over the CoT method consistently, meanwhile refreshing the current stateof-the-art performances.We further demonstrate that our system advances in more faithful, flexible, and explainable logical reasoning.To our knowledge, this is the first to combine symbolic expressions and rules into CoT for logical reasoning with LLMs.Code is open at https://github.com/Aiden0526/SymbCoT.

Misinformation is a prevalent societal issue due to its potential high risks. Out-Of-Context (OOC) misinformation, where authentic images are repurposed with false text, is one of the easiest and most effective ways to mislead audiences. Current methods focus on assessing image- text consistency but lack convincing explanations for their judgments, which are essential for debunking misinformation. While Multimodal Large Language Models (MLLMs) have rich knowledge and innate capability for visual rea- soning and explanation generation, they still lack sophisti- cation in understanding and discovering the subtle cross- modal differences. In this paper, we introduce Sniffer,a novel multimodal large language model specifically engi- neered for OOC misinformation detection and explanation. Snifferemploys two-stage instruction tuning on Instruct- BLIP. The first stage refines the model's concept alignment of generic objects with news-domain entities and the sec- ond stage leverages OOC-specific instruction data gener- ated by language-only GPT-4 to fine-tune the model's dis- criminatory powers. Enhanced by external tools and re- trieval, Sniffernot only detects inconsistencies between text and image but also utilizes external knowledge for con- textual verification. Our experiments show that Sniffersurpasses the original MLLM by over 40% and outperforms state-of-the-art methods in detection accuracy. Snifferalso provides accurate and persuasive explanations as val- idated by quantitative and human evaluations.

Generalizing to out-of-distribution (OOD) data or unseen domain, termed OOD generalization, still lacks appropriate theoretical guarantees. Canonical OOD bounds focus on different distance measurements between source and target domains but fail to consider the optimization property of the learned model. As empirically shown in recent work, sharpness of learned minimum influences OOD generalization. To bridge this gap between optimization and OOD generalization, we study the effect of sharpness on how a model tolerates data change in domain shift which is usually captured by "robustness" in generalization. In this paper, we give a rigorous connection between sharpness and robustness, which gives better OOD guarantees for robust algorithms. It also provides a theoretical backing for "flat minima leads to better OOD generalization". Overall, we propose a sharpness-based OOD generalization bound by taking robustness into consideration, resulting in a tighter bound than non-robust guarantees. Our findings are supported by the experiments on a ridge regression model, as well as the experiments on deep learning classification tasks.

Existing research of video understanding still struggles to achieve in-depth comprehension and reasoning in complex videos, primarily due to the under-exploration of two key bottlenecks: fine-grained spatial-temporal perceptive understanding and cognitive-level video scene comprehension. This paper bridges the gap by presenting a novel solution. We first introduce a novel video Multimodal Large Language Model (MLLM), MotionEpic, which achieves fine-grained pixel-level spatial-temporal video grounding by integrating video spatial-temporal scene graph (STSG) representation. Building upon MotionEpic, we then develop a Video-of-Thought (VoT) reasoning framework. VoT inherits the Chain-of-Thought (CoT) core, breaking down a complex task into simpler and manageable sub-problems, and addressing them step-by-step from a low-level pixel perception to high-level cognitive interpretation. Extensive experiments across various complex video QA benchmarks demonstrate that our overall framework strikingly boosts existing state-of-the-art. To our knowledge, this is the first attempt at successfully implementing the CoT technique for achieving human-level video reasoning, where we show great potential in extending it to a wider range of video understanding scenarios. Systems and codes will be open later.

While existing Aspect-based Sentiment Analysis (ABSA) has received extensive effort and advancement, there are still gaps in defining a more holistic research target seamlessly integrating multimodality, conversation context, fine-granularity, and also covering the changing sentiment dynamics as well as cognitive causal rationales. This paper bridges the gaps by introducing a multimodal conversational ABSA, where two novel subtasks are proposed: 1) Panoptic Sentiment Sextuple Extraction, panoramically recognizing holder, target, aspect, opinion, sentiment, rationale from multi-turn multi-party multimodal dialogue. 2) Sentiment Flipping Analysis, detecting the dynamic sentiment transformation throughout the conversation with the causal reasons. To benchmark the tasks, we construct PanoSent, a dataset annotated both manually and automatically, featuring high quality, large scale, multimodality, multilingualism, multi-scenarios, and covering both implicit&explicit sentiment elements. To effectively address the tasks, we devise a novel Chain-of-Sentiment reasoning framework, together with a novel multimodal large language model (namely Sentica) and a paraphrase-based verification mechanism. Extensive evaluations demonstrate the superiority of our methods over strong baselines, validating the efficacy of all our proposed methods. The work is expected to open up a new era for the ABSA community, and thus all our codes and data are open at https://PanoSent.github.io/.

Class-incremental continual learning is a core step towards developing artificial intelligence systems that can continuously adapt to changes in the environment by learning new concepts without forgetting those previously learned. This is especially needed in the medical domain where continually learning from new incoming data is required to classify an expanded set of diseases. In this work, we focus on how old knowledge can be leveraged to learn new classes without catastrophic forgetting. We propose a framework that comprises of two main components: (1) a dynamic architecture with expanding representations to preserve previously learned features and accommodate new features; and (2) a training procedure alternating between two objectives to balance the learning of new features while maintaining the model’s performance on old classes. Experiment results on multiple medical datasets show that our solution is able to achieve superior performance over state-of-the-art baselines in terms of class accuracy and forgetting.

To replicate the success of text-to-image (T2I) generation, recent works employ large-scale video datasets to train a text-to-video (T2V) generator. Despite their promising results, such paradigm is computationally expensive. In this work, we propose a new T2V generation setting—One-Shot Video Tuning, where only one text-video pair is presented. Our model is built on state-of-the-art T2I diffusion models pre-trained on massive image data. We make two key observations: 1) T2I models can generate still images that represent verb terms; 2) extending T2I models to generate multiple images concurrently exhibits surprisingly good content consistency. To further learn continuous motion, we introduce Tune-A-Video, which involves a tailored spatio-temporal attention mechanism and an efficient one-shot tuning strategy. At inference, we employ DDIM inversion to provide structure guidance for sampling. Extensive qualitative and numerical experiments demonstrate the remarkable ability of our method across various applications.

Humans can watch a continuous video stream and effortlessly perform continual acquisition and transfer of new knowledge with minimal supervision yet retaining previously learnt experiences. In contrast, existing continual learning (CL) methods require fully annotated labels to effectively learn from individual frames in a video stream. Here, we examine a more realistic and challenging problem—Label-Efficient Online Continual Object Detection (LEOCOD) in streaming video. We propose a plug-and-play module, Efficient-CLS, that can be easily inserted into and consistently improve existing CL algorithms for object detection in video streams with reduced data annotation costs and model retraining time. We show that our method has achieved significant improvement with minimal forgetting across all supervision levels on two challenging CL benchmarks for streaming real-world videos. Remarkably, with only 25% annotated video frames, our proposed method still outperforms the state-of-the-art CL models trained with 100% annotations on all video frames. The data and source code will be publicly available at https://github.com/showlab/Efficient-CLS.

Patients with co-morbidities often require multiple medications to manage their conditions. However, existing medication recommendation systems only offer class-level medications and regard all interactions among drugs to have the same level of severity. This limits their ability to provide personalized and safe recommendations tailored to individual needs. In this work, we introduce a deep learning-based fine-grained medication recommendation system called REFINE, which is designed to improve treatment outcomes and minimize adverse drug interactions. In order to better characterize patients’ health conditions, we model the trend in medication dosage titrations and lab test responses, and adapt the vision transformer to obtain effective patient representations. We also model drug interaction severity levels as weighted graphs to learn safe drug combinations and design a balanced loss function to avoid overly conservative recommendations and miss medications that might be needed for certain conditions. Extensive experiments on two real-world datasets show that REFINE outperforms state-of-the-art techniques.

Discovering object-centric representations from images has the potential to greatly improve the robustness, sample efficiency and interpretability of machine learning algorithms. Current works on multi-object images typically follow a generative approach that optimizes for input reconstruction and fail to scale to real-world datasets despite significant increases in model capacity. We address this limitation by proposing a novel method that leverages feature connectivity to cluster neighboring pixels likely to belong to the same object. We further design two object-centric regularization terms to refine object representations in the latent space, enabling our approach to scale to complex real-world images. Experimental results on simulated, real-world, complex texture and common object images demonstrate a substantial improvement in the quality of discovered objects compared to state-of-the-art methods, as well as the sample efficiency and generalizability of our approach. We also show that the discovered object-centric representations can accurately predict key object properties in downstream tasks, highlighting the potential of our method to advance the field of multi-object representation learning.

OBJECTIVE: To develop a deep learning algorithm (DLA) to detect diabetic kideny disease (DKD) from retinal photographs of patients with diabetes, and evaluate performance in multiethnic populations. MATERIALS AND METHODS: We trained 3 models: (1) image-only; (2) risk factor (RF)-only multivariable logistic regression (LR) model adjusted for age, sex, ethnicity, diabetes duration, HbA1c, systolic blood pressure; (3) hybrid multivariable LR model combining RF data and standardized z-scores from image-only model. Data from Singapore Integrated Diabetic Retinopathy Program (SiDRP) were used to develop (6066 participants with diabetes, primary-care-based) and internally validate (5-fold cross-validation) the models. External testing on 2 independent datasets: (1) Singapore Epidemiology of Eye Diseases (SEED) study (1885 participants with diabetes, population-based); (2) Singapore Macroangiopathy and Microvascular Reactivity in Type 2 Diabetes (SMART2D) (439 participants with diabetes, cross-sectional) in Singapore. Supplementary external testing on 2 Caucasian cohorts: (3) Australian Eye and Heart Study (AHES) (460 participants with diabetes, cross-sectional) and (4) Northern Ireland Cohort for the Longitudinal Study of Ageing (NICOLA) (265 participants with diabetes, cross-sectional). RESULTS: In SiDRP validation, area under the curve (AUC) was 0.826(95% CI 0.818-0.833) for image-only, 0.847(0.840-0.854) for RF-only, and 0.866(0.859-0.872) for hybrid. Estimates with SEED were 0.764(0.743-0.785) for image-only, 0.802(0.783-0.822) for RF-only, and 0.828(0.810-0.846) for hybrid. In SMART2D, AUC was 0.726(0.686-0.765) for image-only, 0.701(0.660-0.741) in RF-only, 0.761(0.724-0.797) for hybrid. DISCUSSION AND CONCLUSION: There is potential for DLA using retinal images as a screening adjunct for DKD among individuals with diabetes. This can value-add to existing DLA systems which diagnose diabetic retinopathy from retinal images, facilitating primary screening for DKD.

OBJECTIVE: This study aims to develop a convolutional neural network-based learning framework called domain knowledge-infused convolutional neural network (DK-CNN) for retrieving clinically similar patient and to personalize the prediction of macrovascular complication using the retrieved patients. MATERIALS AND METHODS: We use the electronic health records of 169 434 patients with diabetes, hypertension, and/or lipid disorder. Patients are partitioned into 7 subcohorts based on their comorbidities. DK-CNN integrates both domain knowledge and disease trajectory of patients over multiple visits to retrieve similar patients. We use normalized discounted cumulative gain (nDCG) and macrovascular complication prediction performance to evaluate the effectiveness of DK-CNN compared to state-of-the-art models. Ablation studies are conducted to compare DK-CNN with reduced models that do not use domain knowledge as well as models that do not consider short-term, medium-term, and long-term trajectory over multiple visits. RESULTS: Key findings from this study are: (1) DK-CNN is able to retrieve clinically similar patients and achieves the highest nDCG values in all 7 subcohorts; (2) DK-CNN outperforms other state-of-the-art approaches in terms of complication prediction performance in all 7 subcohorts; and (3) the ablation studies show that the full model achieves the highest nDCG compared with other 2 reduced models. DISCUSSION AND CONCLUSIONS: DK-CNN is a deep learning-based approach which incorporates domain knowledge and patient trajectory data to retrieve clinically similar patients. It can be used to assist physicians who may refer to the outcomes and past treatments of similar patients as a guide for choosing an effective treatment for patients.

Chronic disease management involves frequent administration of invasive lab procedures in order for clinicians to determine the best course of treatment regimes for these patients. However, patients are often put off by these invasive lab procedures and do not follow the appointment schedules. This has resulted in poor management of their chronic conditions leading to unnecessary disease complications. An AI system that is able to personalize the prediction of individual patient lab test responses will enable clinicians to titrate the medications to achieve the desired therapeutic outcome. Accurate prediction of lab test response is a challenge because these patients typically have co-morbidities and their treatments might influence the target lab test response. To address this, we model the complex interactions among different medications, diseases, lab test response, and fine-grained dosage information to learn a strong patient representation. Together with information from similar patients and external knowledge such as drug-lab interactions and diagnosis-lab interaction, we design a system called KALP to perform personalized prediction of patients’ response for a target lab result and identify the top influencing factors for the prediction. Experiment results on real-world datasets demonstrate the effectiveness of KALP in reducing prediction errors by a significant margin. Case studies show that the identified factors are consistent with clinicians’ understanding.

Predicting disease progression is key to provide stratified patient care and enable good utilization of healthcare resources. The availability of longitudinal images has enabled image-based disease progression prediction. In this work, we propose a framework called DP-GAT to identify regions containing significant biological structures and model the relationships among these regions as a graph along with their respective contexts. We perform reasoning via Graph Attention Network to generate representations that enable accurate disease progression prediction. We further extend DP-GAT to perform 3D medical volume segmentation. Experiments on real world medical image datasets demonstrate the advantage of our approach over strong baseline methods for both disease progression prediction and 3D segmentation tasks.

The broad adoption of electronic health records (EHRs) has led to vast amounts of data being accumulated on a patient’s history, diagnosis, prescriptions, and lab tests. Advances in recommender technologies have the potential to utilize this information to help doctors personalize the prescribed medications. However, existing medication recommendation systems have yet to make use of all these information sources in a seamless manner, and they do not provide a justification on why a particular medication is recommended. In this work, we design a two-stage personalized medication recommender system called PREMIER that incorporates information from the EHR. We utilize the various weights in the system to compute the contributions from the information sources for the recommended medications. Our system models the drug interaction from an external drug database and the drug co-occurrence from the EHR as graphs. Experiment results on MIMIC-III and a proprietary outpatient dataset show that PREMIER outperforms state-of-the-art medication recommendation systems while achieving the best tradeoff between accuracy and drug-drug interaction. Case studies demonstrate that the justifications provided by PREMIER are appropriate and aligned to clinical practices.

Deep learning-based models are developed to automatically detect if a retina image is ‘referable’ in diabetic retinopathy (DR) screening. However, their classification accuracy degrades as the input images distributionally shift from their training distribution. Further, even if the input is not a retina image, a standard DR classifier produces a high confident prediction that the image is ‘referable’. Our paper presents a Dirichlet Prior Network-based framework to address this issue. It utilizes an out-of-distribution (OOD) detector model and a DR classification model to improve generalizability by identifying OOD images. Experiments on real-world datasets indicate that the proposed framework can eliminate the unknown non-retina images and identify the distributionally shifted retina images for human intervention.

Clinical applications that involves risk stratification of patients often predict the likely occurrence of an event such as the development of a complication as well as the time to the next event. Existing approaches that provide individual survival distribution across time often use baseline measurements for the risk prediction of an event occurring, and is unable to estimate the time to next event. Further, they do not deal with the complex and heterogeneous information from past visit records, and hence do not handle repeated events which are common in chronic diseases. We address the above limitations by designing a recurrent temporal time process network that incorporates longitudinal visit information to increase the accuracy of risk predictions for clinical events and to estimate the time to next event. Our proposed solution utilizes a recurrent neural network to learn a latent representation of patient visit history. This representation enables us to approximate the conditional intensity function of the temporal point process, from which we derive the survival probability function to obtain the risk of an event occurring and to estimate the time to next event. We demonstrate our approach on two real-world healthcare datasets and show that the proposed approach is able to achieve a significant performance improvement over state-of-the-art methods.

Data augmentation has improved the accuracy and robustness of deep neural networks. Research has focused on finding an optimal augmentation policy that generates good quality training images to improve classification accuracy. However, searching for this optimal augmentation policy is computationally expensive and is dependent on the neural architecture. In this work, we design a dynamic augmentation approach that automatically adjusts the number of transformation operations and their magnitudes during the training of deep neural networks. We also address the shift in the test data distribution by proposing to perform augmentation on the test data. We validate the effectiveness of our solution on CIFAR-10, CIFAR-100, ImageNet, and the perturbed datasets including CIFAR-10-C, CIFAR-100-C, ImageNet-A, ImageNet-C and ImageNet-P. Experiment results show that our proposed dynamic augmentation approach is scalable and gives good performances on clean, adversarial and corrupt datasets, reducing the best published results by a significant margin.

Learning concepts that are consistent with human perception is important for Deep Neural Networks to win end-user trust. Post-hoc interpretation methods lack transparency in the feature representations learned by the models. This work proposes a guided learning approach with an additional concept layer in a CNN-based architecture to learn the associations between visual features and word phrases. We design an objective function that optimizes both prediction accuracy and semantics of the learned feature representations. Experiment results demonstrate that the proposed model can learn concepts that are consistent with human perception and their corresponding contributions to the model decision without compromising accuracy. Further, these learned concepts are transferable to new classes of objects that have similar concepts.

Existing works have generated explanations for deep neural network decisions to provide insights into model behavior. We observe that these explanations can also be used to identify concepts that caused misclassifications. This allows us to understand the possible limitations of the dataset used to train the model, particularly the under-represented regions in the dataset. This work proposes a framework that utilizes concept-based explanations to automatically augment the dataset with new images that can cover these under-represented regions to improve the model performance. The framework is able to use the explanations generated by both interpretable classifiers and post-hoc explanations from black-box classifiers. Experiment results demonstrate that the proposed approach improves the accuracy of classifiers compared to state-of-the-art augmentation strategies.

Although deep learning for Diabetic Retinopathy (DR) screening has shown great success in achieving clinically acceptable accuracy for referable versus non-referable DR, there remains a need to provide more fine-grained grading of the DR severity level as well as automated segmentation of lesions (if any) in the retina images. We observe that the DR severity level of an image is dependent on the presence of different types of lesions and their prevalence. In this work, we adopt a multi-task learning approach to perform the DR grading and lesion segmentation tasks. In light of the lack of lesion segmentation mask ground-truths, we further propose a semi-supervised learning process to obtain the segmentation masks for the various datasets. Experiments results on publicly available datasets and a real world dataset obtained from population screening demonstrate the effectiveness of the multi-task solution over state-of-the-art networks.

Bayesian approaches have been successfully applied in social network analysis to study group behaviors such as online information dissemination and voting pattern. The focus has been on estimating the structure and strength of peer influence and its impact on the decisions of an individual. Less attention has been given to incorporating contextual information and individuals' hidden characteristics (or bias). In this work, we examine the social dynamics where social influence and contextual information play pivotal roles in driving one's decision. We design a probabilistic graphical model called CLAP to understand users' decision behavior in a social network, with an emphasis on both social-level and individual-level factors. To this end, the proposed model introduces hidden bias states associated with each actor and jointly estimates each actor's hidden bias state together with the social influence network. We demonstrate the effectiveness of CLAP on two types of social networks, a real-world US Congress network where senators vote on new bills, and online Twitter networks where users debate on the effectiveness of vaccine and lockdown policy during COVID-19. The experiment results show that CLAP outperforms state-of-the-art game theoretic approaches in predicting user decision. Further, the estimated social influence networks by CLAP has high edge homogeniety ratios.

A fundamental factor limiting the effectiveness of classification algorithms, especially in the medical imaging domain, has been an insufficient quantity of relevant class-specific data. In particular, positive examples of disease conditions tend to be rare, and represent a common bottleneck in improving model performance. In this paper, we introduce GAN-based generative data augmentation methods with dynamic input sampling, and compare their performance against an image feature transfer technique, towards improving the performance of real-world diabetic retinopathy classification tasks. Results suggest that generative data augmentation has the potential to significantly improve classification performance over the baseline.

Passage retrieval is a part of fact-checking and question answering systems that is critical yet often neglected. Most systems usually rely only on traditional sparse retrieval. This can have a significant impact on the recall, especially when the relevant passages have few overlapping words with the query sentence. Recent approaches have attempted to learn dense representations of queries and passages to better capture the latent semantic content of text. While dense retrieval models have been proven effective in question answering, there is no relevant work for improving evidence retrieval in fact-checking. In this work, we show that training a dense retriever is sufficient to outperform traditional sparse representations in both question answering and fact-checking. We constructed a new dataset called Factual-NLI, comprised of factual claims and their supporting evidence, and demonstrate that using it to train a dense retriever can improve evidence retrieval significantly. Experimental results on the MSMARCO dataset indicate that pre-training with Factual-NLI, and other NLI datasets, is also effective for large-scale passage retrieval in question answering. Our model is incorporated in a real world semantic search engine that returns snippets containing evidence related to questions and claims about the COVID-19 pandemic.

Existing defenses against adversarial attacks are typically tailored to a specific perturbation type. Using adversarial training to defend against multiple types of perturbation requires expensive adversarial examples from different perturbation types at each training step. In contrast, manifold-based defense incorporates a generative network to project an input sample onto the clean data manifold. This approach eliminates the need to generate expensive adversarial examples while achieving robustness against multiple perturbation types. However, the success of this approach relies on whether the generative network can capture the complete clean data manifold, which remains an open problem for complex input domain. In this work, we devise an approximate manifold defense mechanism, called RBF-CNN, for image classification. Instead of capturing the complete data manifold, we use an RBF layer to learn the density of small image patches. RBF-CNN also utilizes a reconstruction layer that mitigates any minor adversarial perturbations. Further, incorporating our proposed reconstruction process for training improves the adversarial robustness of our RBF-CNN models. Experiment results on MNIST and CIFAR-10 datasets indicate that RBF-CNN offers robustness for multiple perturbations without the need for expensive adversarial training.

Among existing uncertainty estimation approaches, Dirichlet Prior Network (DPN) distinctly models different predictive uncertainty types. However, for in-domain examples with high data uncertainties among multiple classes, even a DPN model often produces indistinguishable representations from the out-of-distribution (OOD) examples, compromising their OOD detection performance. We address this shortcoming by proposing a novel loss function for DPN to maximize the representation gap between in-domain and OOD examples. Experimental results demonstrate that our proposed approach consistently improves OOD detection performance.

Though deep learning systems have achieved high accuracy in detecting diseases from medical images, few such systems have been deployed in highly automated disease screening settings due to lack of trust in how well these systems can generalize to out-of-datasets. We propose to use uncertainty estimates of the deep learning system’s prediction to know when to accept or to disregard its prediction. We evaluate the effectiveness of using such estimates in a real-life application for the screening of diabetic retinopathy. We also generate visual explanation of the deep learning system to convey the pixels in the image that influences its decision. Together, these reveal the deep learning system’s competency and limits to the human, and in turn the human can know when to trust the deep learning system.

Ischemic stroke is a leading cause of death and long-term disability that is difficult to predict reliably. Retinal fundus photography has been proposed for stroke risk assessment, due to its non-invasiveness and the similarity between retinal and cerebral microcirculations, with past studies claiming a correlation between venular caliber and stroke risk. However, it may be that other retinal features are more appropriate. In this paper, extensive experiments with deep learning on six retinal datasets are described. Feature isolation involving segmented vascular tree images is applied to establish the effectiveness of vessel caliber and shape alone for stroke classification, and dataset ablation is applied to investigate model generalizability on unseen sources. The results suggest that vessel caliber and shape could be indicative of ischemic stroke, and sourcespecific features could influence model performance.

Explaining the decisions of a Deep Learning Network is imperative to safeguard end-user trust. Such explanations must be intuitive, descriptive, and faithfully explain why a model makes its decisions. In this work, we propose a framework called FLEX (Faithful Linguistic EXplanations) that generates post-hoc linguistic justifications to rationalize the decision of a Convolutional Neural Network. FLEX explains a model’s decision in terms of features that are responsible for the decision. We derive a novel way to associate such features to words, and introduce a new decision-relevance metric that measures the faithfulness of an explanation to a model’s reasoning. Experiment results on two benchmark datasets demonstrate that the proposed framework can generate discriminative and faithful explanations compared to state-of-the-art explanation generators. We also show how FLEX can generate explanations for images of unseen classes as well as automatically annotate objects in images.

Recent deep neural networks (DNN) utilize identity mappings involving either element-wise addition or channel-wise concatenation for the propagation of these identity mappings. In this paper, we propose a new propagation mechanism called channel-wise addition (cAdd) to deal with the vanishing gradients problem without sacrificing the complexity of the learned features. Unlike channel-wise concatenation, cAdd is able to eliminate the need to store feature maps thus reducing the memory requirement. The proposed cAdd mechanism can deepen and widen existing neural architectures with fewer parameters compared to channel-wise concatenation and element-wise addition. We incorporate cAdd into state-of-the-art architectures such as ResNet, WideResNet, and CondenseNet and carry out extensive experiments on CIFAR10, CIFAR100, SVHN and ImageNet to demonstrate that cAdd-based architectures are able to achieve much higher accuracy with fewer parameters compared to their corresponding base architectures.

Over-fitting is a common issue of training deep convolutional neural network especially when the dataset is limited. In this work, we propose a patch-level regularizer to force a convolutional neural network to learn many sub-models during training, and aggregate these models during testing. This approach proves to be robust and noise tolerant as our regularizer exposes small patches of an image to the network to learn all the features equally during training. The regularizer can be easily applied to the convolutional neural networks to further improve their classification performance. Experiment results on publicly available datasets demonstrate consistent improvement over existing regularizers.

Detection of Adverse Drug Events (ADE) or side effects of different treatments are necessary to minimize potential health risks of patients. Given the prevalence of user reported content on the web, recent research has focused on the automatic discovery of potential side effects from these online platforms. However, it is not clear whether the symptoms that a patient experiences, are solely side effects of a particular treatment (or a combination of them), or there are other confounding factors influencing them. In this work, we characterize the reported symptoms along with their severity for patients, based on their past interactions with various treatments and their pre-existing medical conditions. We analyze a large dataset from a symptoms tracking app, and observe a strong correlation between a patient's existing health condition(s) and the symptoms he or she experiences across different treatments. We develop a multi-objective neural network with gating mechanism, to predict the possible symptoms and their overall severity level for a set of treatments, for a given patient. Experimental results demonstrate the effectiveness of our model over state-of-the-art approaches. Furthermore, our adaptation of the gating mechanism imbues the network with the ability of justifying its predictions.

Gaussian distributions are commonly used as a key building block in many generative models. However, their applicability has not been well explored in deep networks. In this paper, we propose a novel deep generative model named as Normal Similarity Network (NSN) where the layers are constructed with Gaussian-style filters. NSN is trained with a layer-wise non-parametric density estimation algorithm that iteratively down-samples the training images and capture the density of the down-sampled training images in the final layer. Additionally, we propose NSN-Gen for generating new samples from noise vectors by iteratively reconstructing feature maps in the hidden layers of NSN. Our experiments suggest encouraging results of the proposed model for a wide range of computer vision applications including image generation, styling and reconstruction from occluded images.

Vessel type classification is a preliminary step in quantifying the severity of various diseases. This paper proposes DBA, a simple yet effective vessel type classification method based on the principle that arteries are brighter than veins at the local scale. The weighted local difference of the red channel intensity of the main trunk of each vessel is compared with that of its two immediately neighbouring vessels, a feature that is highly correlated with vessel-rectified oxygen capacity, and in turn, vessel type. Experiments on the publicly-available INSPIRE-AVR and DRIVE datasets obtained average vessel accuracies of 0.9217/0.9071, and average pixel accuracies of 0.9602/0.9634 respectively, with particular effectiveness on images with low contrast, non-uniform illumination and colour variation confirmed on the SiMES 1 dataset.

Detecting rumors is a crucial task requiring significant time and manual effort in forms of investigative journalism. In social media such as Twitter, unverified information can get disseminated rapidly making early detection of potentially false rumors critical. We observe that the early reactions of people towards an emerging claim can be predictive of its veracity. We propose a novel neural network architecture using the stances of people engaging in a conversation on Twitter about a rumor for detecting its veracity. Our proposed solution comprises two key steps. We first detect the stance of each individual tweet, by considering the textual content of the tweet, its timestamp, as well as the sequential conversation structure leading up to the target tweet. Then we use the predicted stances of all tweets in a conversation tree to determine the veracity of the original rumor. We evaluate our model on the SemEval2017 rumor detection dataset and demonstrate that our solution outperforms the state-of-the-art approaches for both stance prediction and rumor veracity prediction tasks.

I am very happy to report that TSC has gained an Impact Factor (IF) of 3.520 and the 5-year IF of 4.245, both of which represent significant increases from the previous years. This further speaks to the global reputation of the journal and the amazing work done by the past EICs, all the current and past EB members, and reviewers - all of whom have volunteered their precious time despite their very busy schedule to support and contribute to the growth of this journal. I hope to count on your continued engagement for the future growth of this journal. Over this past year, several esteemed EB members have completed their terms of service to TSC after serving for several years. On behalf of the Services Computing community and the TSC EAB, I would like to thank the following Associate Editors who retired from TSC EB in 2017 for their invaluable service and contributions to the journal. Overall, I am very proud of the success that TSC has achieved in 2017. This would not have been possible without the continued support of the authors, readers, reviewers, TSC EAB, TSC EB, and the staff of IEEE and IEEE Computer Society. I look forward to exploring ways to further enhance the reputation and impact of our journal. I would love to hear your suggestions and comments, and I hope to have your continued support.

Posts by users on microblogs such as Twitter provide diverse real-time updates to major events. Unfortunately, not all the information are credible. Previous works that assess the credibility of information in Twitter have focused on extracting features from the Tweets. In this work, we present an interactive framework called iFACT for assessing the credibility of claims from tweets. The proposed framework collects independent evidence from web search results (WSR) and identify the dependencies between claims. It utilizes features from the search results to determine the probabilities that a claim is credible, not credible or inconclusive. Finally, the dependencies between claims are used to adjust the likelihood estimates of a claim being credible, not credible or inconclusive. iFACT allows users to be engaged in the credibility assessment process by providing feedback as to whether the web search results are relevant, support or contradict a claim. Experiment results on multiple real world datasets demonstrate the effectiveness of WSR features and its ability to generalize to claims of new events. Case studies show the usefulness of claim dependencies and how the proposed approach can give explanation to the credibility assessment process.

User generated content about products and services in the form of reviews are often diverse and even contradictory.This makes it difficult for users to know if an opinion in a review is prevalent or biased.We study the problem of searching for supporting opinions in the context of reviews.We propose a framework called SURF, that first identifies opinions expressed in a review, and then finds similar opinions from other reviews.We design a novel probabilistic graphical model that captures opinions as a combination of aspect, topic and sentiment dimensions, takes into account the preferences of individual authors, as well as the quality of the entity under review, and encodes the flow of thoughts in a review by constraining the aspect distribution dynamically among successive review segments.We derive a similarity measure that considers both lexical and semantic similarity to find supporting opinions.Experiments on TripAdvisor hotel reviews and Yelp restaurant reviews show that our model outperforms existing methods for modeling opinions, and the proposed framework is effective in finding supporting opinions.

In this demonstration, we showcase MAROON+, a system that builds historical profiles of real-world target entities by integrating their publicly available information from different Web sites. We face two challenges when building such a system. First, an entity may change its attribute values over time and it is thus hard to decide if distinct values actually describe the same entity but at different times. Second, the Web sites may provide inaccurate information or fail to update their contents in a timely manner. MAROON+ employs a source-aware matching algorithm that jointly considers the evolution of entities and the source quality to link temporal records to a target entity. We characterize a source by its precision which measures the probability that a published value conforms to the real world, recall which measures the probability that a change in real world is captured by a source, and freshness which measures the timeliness of a published value.

The diffusion of rumors is a major concern for web users. Limiting the spread of rumor on social networks has become an important task. One approach is to identify nodes to start a truth campaign such that when users are aware of the truth, they would not believe or propagate the rumor. However, existing works do not take into account the delays of information diffusion or the time point beyond which propagation of misinformation is no longer critical. In this paper, we consider a more realistic situation where information is propagated with delays and the goal is to reduce the number of rumor-infected users before a deadline. We call this the Temporal Influence Blocking (TIB) problem. We propose a two-phase solution called TIB-Solver to select k nodes to start a truth campaign such that the number of users reached by a rumor is minimized. Experiments show that the proposed TIBSolver outperforms the state-of-the-art algorithms in terms of both effectiveness and efficiency.

User opinions expressed in the form of ratings can influence an individual's view of an item. However, the true quality of an item is often obfuscated by user biases, and it is not obvious from the observed ratings the importance different users place on different aspects of an item. We propose a probabilistic modeling of the observed aspect ratings to infer (i) each user's aspect bias and (ii) latent intrinsic quality of an item. We model multi-aspect ratings as ordered discrete data and encode the dependency between different aspects by using a latent Gaussian structure. We handle the Gaussian-Categorical non-conjugacy using a stick-breaking formulation coupled with P\'{o}lya-Gamma auxiliary variable augmentation for a simple, fully Bayesian inference. On two real world datasets, we demonstrate the predictive ability of our model and its effectiveness in learning explainable user biases to provide insights towards a more reliable product quality estimation.

To harness the rich amount of information available on the web today, many organizations aggregate public (and private) data to derive knowledge repositories for real-world entities. This paper aims to build historical profiles of real-world entities by integrating temporal records collected from different sources. This problem is challenging not only because entities may change their attribute values over time, but also because information provided by the sources could be unreliable. In this paper, we present a new solution for profiling entities over time. To understand the evolution of entities, we describe a novel transition model which gives the probability that an entity will change to a particular attribute value after some time period. Next, a set of quality metrics are defined for the data sources to capture the exactness and timeliness of their provided values. The transition model and the quality metrics are then built into a source-aware temporal matching algorithm that can link temporal records to entities at the right time and augment entity profiles with correct values. Our suite of experiments demonstrate that the proposed approach is able to outperform the state-of-the-art techniques by constructing more complete and accurate profiles for entities.

Influence maximization (IM) problem asks for a set of k nodes in a given graph G, such that it can reach the largest expected number of remaining nodes in G. Existing methods have either considered that the influence be targeted to meet certain deadline constraint, or be restricted to specific geographical region. However, if an event organizer wants to disseminate some event information on a social platform, s/he would want to select a set of users who can influence the most number of people within the neighborhood of the event location, and this influence should occur before the event takes place. Considering the location and deadline independently may lead to a less than optimal set of users. In this paper, we formalize the problem targeted influence maximization in social networks. We adopt a login model where each user is associated with a login probability and he can be influenced by his neighbors only when he is online. We develop a sampling based algorithm that returns a (1-1/e-ε)-approximate solution, as well as an efficient heuristic algorithm that focuses on nodes close to the target location. Experiments on real-world social network datasets demonstrate the effectiveness and efficiency of our proposed method.

Diabetic retinopathy (DR) might be characterized by the occurrence of lesions in the retinal image. Existing approaches require a large set of retinal images where lesions in the image are individually annotated to learn a model that will classify an image as referable or non-referable DR. However, annotating individual lesions is a tedious task and the accuracy of the learnt model is limited by the availability of these annotated images. In this paper, we first learn a universal Gaussian mixture model (GMM) from a small set of annotated images. This universal GMM is then applied as the prior belief to learn an adaptive GMM for individual images. The proposed approach aims to capture the characteristics of referable versus non-referable images by examining the difference between the universal GMM and the adaptive GMM. An image-level classifier is then built based on these differences as features. Experimental results on three fundus image datasets (MESSIDOR, DIARETDB1 and SORC) indicate that the proposed framework achieves 92.1%, 97.68% and 87.1% ROC area values respectively. This approach also opens up a way to use the widely available public fundus images, where the images are labelled but not annotated, for progressively refining the universal GMM leading to an improved performance of approximately 5% and 1% respectively for SORC and MESSIDOR dataset after five refinement steps.

The theory of brokerage in sociology suggests if contacts between two parties are enabled through a third party, the latter occupies a strategic position of controlling information flows. Such individuals are called brokers and they play a key role in disseminating information. However, there is no systematic approach to identify brokers in online social networks. In this paper, we formally define the problem of detecting top-$k$ brokers given a social network and show that it is NP-hard. We develop a heuristic algorithm to find these brokers based on the weak tie theory. In order to handle the dynamic nature of online social networks, we design incremental algorithms: WeakTie-Local for unidirectional networks and WeakTie-Bi for bidirectional networks. We use two real world datasets, DBLP and Twitter, to evaluate the proposed methods. We also demonstrate how the detected brokers are useful in diffusing information across communities and propagating tweets to reach more distinct users.

Locating nodes to immunize in computer/social networks to control the spread of virus or rumors has become an important problem. In real world contagions, nodes may get infected by external sources when the propagation is underway. While most studies formalize the problem in a setting where contagion starts at one time point, we model a more realistic situation where there are likely to be many breakouts of contagions over a time window. We call this the node immunization over infectious period (NIIP) problem. We show that the NIIP problem is NP-hard and remains so even in directed acyclic graphs. We propose a NIIP algorithm to select $k$ nodes to immunize over a time period. Simulation is performed to estimate a good distribution of $k$ over the time period. For each time point, the NIIP algorithm will make decisions which nodes to immunize given the estimated value of $k$ for that time point. Experiments show that the proposed NIIP algorithm outperform the state-of-the-art algorithms in terms of both effectiveness and efficiency.

Glaucoma is a widespread ocular disorder leading to irreversible loss of vision. Therefore, there is a pressing need for cost-effective screening, such that preventive measures can be taken. This can be achieved with an accurate segmentation of the optic disc and cup from retinal images to obtain the cup-to-disc ratio. We describe a comprehensive solution based on applying convolutional neural networks to feature exaggerated inputs emphasizing disc pallor without blood vessel obstruction, as well as the degree of vessel kinking. The produced raw probability maps then undergo a robust refinement procedure that takes into account prior knowledge about retinal structures. Analysis of these probability maps further allows us to obtain a confidence estimate on the correctness of the segmentation, which can be used to direct the most challenging cases for manual inspection. Tests on two large real-world databases, including the publicly-available MESSIDOR collection, demonstrate the effectiveness of our proposed system.

To harness the rich amount of information available on the Web today, many organizations start to aggregate public (and private) data to derive new knowledge bases. A fundamental challenge in constructing an accurate integrated knowledge repository from different data sources is to understand how facts across different sources are related to one another over time. This challenge, referred to as the temporal record linkage problem, goes far beyond the traditional record linkage problem as it requires a fine-grained analysis of how two facts are temporally related if they both refer to the same entity.

The rapid growth of information sources on the Web has intensified the problem of data quality. In particular, the same real world entity may be described by different sources in various ways with overlapping information, and possibly conflicting or even erroneous values. In order to obtain a more complete and accurate picture for a real world entity, we need to collate the data records that refer to the entity, as well as correct any erroneous values. We observe that these two tasks are often tightly coupled: rectifying erroneous values will facilitate data collation, while linking similar records provides us with a clearer view of the data and additional evidence for error correction. In this paper, we present a framework called Comet that interleaves record linkage with error correction, taking into consideration the source reliabilities on various attributes. The proposed framework first utilizes confidence based matching to discriminate records in terms of ambiguity and source reliability. Then it performs adaptive matching to reduce the impact of erroneous values. Experiment results demonstrate that Comet outperforms the state-of-the-art techniques and is able to build complete and accurate profiles for real world entities.

Advances in Web 2.0 technology has led to the rising popularity of many social network services. For example, there are over 500 million active users in Twitter. Given the huge number of users, user recommendation has gained importance where the goal is to find a set of users whom a target user is likely to follow. Content-based approaches that rely on tweet content for user recommendation have low precision as tweet contents are typically short and noisy, while collaborative filtering approaches that utilize follower-followee relationships lead to higher precision but data sparsity remains a challenge. In this work, we propose a community-based approach to user recommendation in Twitter-style social networks. Forming communities enables us to reduce data sparsity as the focus is on discover the latent characteristics of communities instead of individuals. We employ an LDA-based method on the follower-followee relationships to discover communities before applying the state-of-the-art matrix factorization method on each of the communities. This approach proves effective in improving the conversion rate (by as much as 20%) as demonstrated by the results of extensive experiments on two real world data sets Twitter and Weibo. In addition, the community-based approach is scalable as the individual community can be analyzed separately.

Existing recommendation algorithms assume that users make their purchase decisions solely based on individual preferences, without regard to the type of users nor the products' maturity stages. Yet, extensive studies have shown that there are two types of users: innovators and imitators. Innovators tend to make purchase decisions based solely on their own preferences; whereas imitators' purchase decisions are often influenced by a product's stage of maturity. In this paper, we propose a framework that seamlessly incorporates the type of user and product maturity into existing recommendation algorithms. We apply Bass model to classify each user as either an innovator or imitator according to his/her previous purchase behavior. In addition, we introduce the concept of tipping point of a user. This tipping point refers to the point on the product maturity curve beyond which the user is likely to be more receptive to purchasing the product. We refine two widely-adopted recommendation algorithms to incorporate the effect of product maturity in relation to the user type. Experiment results on a real-world dataset obtained from an E-commerce website show that the proposed approach outperforms existing algorithms.

Given the vast amount of information on the World Wide Web, recommender systems are increasingly being used to help filter irrelevant data and suggest information that would interest users. Traditional systems make recommendations based on a single domain e.g., movie or book domain. Recent work has examined the correlations in different domains and designed models that exploit user preferences on a source domain to predict user preferences on a target domain. However, these methods are based on matrix factorization and can only be applied to two-dimensional data. Transferring high dimensional data from one domain to another requires decomposing the high dimensional data to binary relations which results in information loss.

Existing recommender systems model user interests and the social influences independently. In reality, user interests may change over time, and as the interests change, new friends may be added while old friends grow apart and the new friendships formed may cause further interests change. This complex interaction requires the joint modeling of user interest and social relationships over time. In this paper, we propose a probabilistic generative model, called Receptiveness over Time Model (RTM), to capture this interaction. We design a Gibbs sampling algorithm to learn the receptiveness and interest distributions among users over time. The results of experiments on a real world dataset demonstrate that RTM-based recommendation outperforms the state-of-the-art recommendation methods. Case studies also show that RTM is able to discover the user interest shift and receptiveness change over time

Personalized recommender systems aim to push only the relevant items and information directly to the users without requiring them to browse through millions of web resources. The challenge of these systems is to achieve a high user acceptance rate on their recommendations. In this paper, we aim to increase the user acceptance of recommendations by providing more intuitive tag-based explanations of why the items are recommended. Tags are used as intermediary entities that not only relate target users to the recommended items but also understand users' intents. Our system also allows tag-based online relevance feedback. Experiment results on the Movielens dataset show that the proposed approach is able to increase the acceptance rate of recommendations and improve user satisfaction.

Many frequent pattern mining algorithms have been developed for categorical, numerical, time series, or interval data. However, little attention has been given to integrate these algorithms so as to mine frequent patterns from multiple domain datasets for classification. In this paper, we introduce the notion of a heterogenous pattern to capture the associations among different kinds of data. We propose a unified framework for mining multiple domain datasets and design an iterative algorithm called HTMiner. HTMiner discovers essential heterogenous patterns for classification and performs instance elimination. This instance elimination step reduces the problem size progressively by removing training instances which are correctly covered by the discovered essential heterogenous pattern. Experiments on two real world datasets show that the HTMiner is efficient and can significantly improve the classification accuracy.

Trajectory classification has many useful applications. Existing works on trajectory classification do not consider the duration information of trajectory. In this paper, we extract duration-aware features from trajectories to build a classifier. Our method utilizes information theory to obtain regions where the trajectories have similar speeds and directions. Further, trajectories are summarized into a network based on the MDL principle that takes into account the duration difference among trajectories of different classes. A graph traversal is performed on this trajectory network to obtain the top-k covering path rules for each trajectory. Based on the discovered regions and top-k path rules, we build a classifier to predict the class labels of new trajectories. Experiment results on real-world datasets show that the proposed duration-aware classifier can obtain higher classification accuracy than the state-of-the-art trajectory classifier.

The development of Web 2.0 technology has led to huge economic benefits and challenges for both e-commerce websites and online shoppers. One core technology to increase sales and consumers' satisfaction is the use of recommender systems. Existing product recommender systems consider the order of items purchased by users to obtain a list of recommended items. However, they do not consider the time interval between the products purchased. For example, there is often an interval of 2-3 months between the purchase of printer ink cartridges or refills. Thus, recommending appropriate ink cartridges one week before the user needs to replace the depleted ink cartridges would increase the likelihood of a purchase decision. In this paper, we propose to utilize the purchase interval information to improve the performance of the recommender systems for e-commerce. We design an efficient algorithm to compute the purchase intervals between product pairs from users' purchase history and integrate this information into the marginal utility model. We evaluate our approach on a real world ecommerce dataset. Experimental results demonstrate that our approach significantly improves the conversion rate and temporal diversity compared to state-of-the-art algorithms.

The MaxBRNN problem finds a region such that setting up a new service site within this region would guarantee the maximum number of customers by proximity. This problem assumes that each customer only uses the service provided by his/her nearest service site. However, in reality, a customer tends to go to his/her k nearest service sites. To handle this, MaxBRNN can be extended to the MaxBRkNN problem which finds an optimal region such that setting up a service site in this region guarantees the maximum number of customers who would consider the site as one of their k nearest service locations. We further generalize the MaxBRkNN problem to reflect the real world scenario where customers may have different preferences for different service sites, and at the same time, service sites may have preferred targeted customers. In this paper, we present an efficient solution called MaxFirst to solve this generalized MaxBRkNN problem. The algorithm works by partitioning the space into quadrants and searches only in those quadrants that potentially contain an optimal region. During the space partitioning, we compute the upper and lower bounds of the size of a quadrant's BRkNN, and use these bounds to prune the unpromising quadrants. Experiment results show that MaxFirst can be two to three orders of magnitude faster than the state-of-the-art algorithm.

In this paper we present a distributed reinforcement learning system that leverages on expert coordination knowledge to improve learning in multi-agent problems. We focus on the scenario where agents can communicate with their neighbors but this communication structure and the number of agents may change over time. We express coordination knowledge as constraints to reduce the joint action space for exploration. We introduce an extra learning level to learn when to make use of these constraints. This extra level is decentralized among the agents, making it suitable for our communication restrictions. Experiment results on tactical real-time strategy and soccer games show that our system is effective in online learning as opposed to existing methods that use individual constraints on agents and coordinated action selection.

Influenza viruses mutate frequently and new mutations may emerge while old mutations disappear over a period of time. In addition, some mutations may be dominant in one sub-population but not in the other. Discovering such mutations can help to customize vaccines to increase the effectiveness for targeted group of people. In this paper, we study the problem of mining discriminative mutation chains from two influenza A virus protein datasets, D1 and D2, such that the mutations are frequent and significant in one dataset but infrequent and insignificant in the other dataset. We present an efficient algorithm called DMMiner to discover discriminative mutation chains. Experiments results on the real world influenza A virus protein datasets reveal that DMMiner is able to find interesting discriminative mutation chains involving the H1N1 2009 influenza A virus as well as region-specific mutations involving H5N1.

Social network systems such as FaceBook and YouTube have played a significant role in capturing both explicit and implicit user preferences for different items in the form of ratings and tags. This forms a quaternary relationship among users, items, tags and ratings. Existing systems have utilized only ternary relationships such as users-items-ratings, or users-items-tags to derive their recommendations. In this paper, we show that ternary relationships are insufficient to provide accurate recommendations. Instead, we model the quaternary relationship among users, items, tags and ratings as a 4-order tensor and cast the recommendation problem as a multi-way latent semantic analysis problem. A unified framework for user recommendation, item recommendation, tag recommendation and item rating prediction is proposed. The results of extensive experiments performed on a real world dataset demonstrate that our unified framework outperforms the state-of-the-art techniques in all the four recommendation tasks.

The increasing infectious disease outbreaks has led to a need for new research to better understand the disease's origins, epidemiological features and pathogenicity caused by fast-mutating, fast-spreading viruses. Traditional sequence analysis methods do not take into account the spatio-temporal dynamics of rapidly evolving and spreading viral species. They are also focused on identifying single-point mutations. In this paper, we propose a novel approach that incorporates space-time relationships for studying changes in protein sequences from fast mutating viruses. We aim to detect both single-point mutations as well as k-mutations in the viral sequences. We define the problem of mutation chain pattern mining and design algorithms to discover valid mutation chains. Compact data structures to facilitate the mining process as well as pruning strategies to increase the scalability of the algorithms are devised. Experiments on both synthetic datasets and real world influenza A virus dataset show that our algorithms are scalable and effective in discovering mutations that occur geographically over time.

In an era of increasingly complex biological datasets, one of the key steps in gene functional analysis comes from clustering genes based on co-expression. Biclustering algorithms can identify gene clusters with local co-expressed patterns, which are more likely to define genes functioning together than global clustering methods. However, these algorithms are not effective in uncovering gene regulatory networks because the mined biclusters lack genes that may be critical in the function but may not be co-expressed with the clustered genes. In this paper, we introduce a biclustering method called skeleton biclustering (SKB), which builds high quality biclusters from microarray data, creates relationships among the biclustered genes based on gene ontology annotations, and identifies genes that are missing in the biclusters. SKB thus defines inter-bicluster and intra-bicluster functional relationships. The delineation of functional relationships and incorporation of such missing genes may help biologists to discover biological processes that are important in a given study and provides clues for how the processes may be functioning together. Experimental results show that, with SKB, the biological significance of the biclusters is considerably improved.

Compared to relational data models, the hierarchical structure of semi-structured data such as XML provides semantically meaningful neighbourhoods advancing data cleaning problems such as outlier detection. In this paper, we introduce the concept of correlated subspace that leverages on the hierarchical relationships between XML attributes to provide contextually informative neighbourhoods for attribute outlier detection. We also design two correlation-based attribute outlier metrics for XML, namely the xO-Measure and xQ-Measure. The effectiveness of our XML outlier detection approach is supported with experimental results.

Existing temporal pattern mining assumes that events do not have any duration. However, events in many real world applications have durations, and the relationships among these events are often complex. These relationships are modeled using a hierarchical representation that extends Allen's interval algebra. However, this representation is lossy as the exact relationships among the events cannot be fully recovered. In this paper, we augment the hierarchical representation with additional information to achieve a lossless representation. An efficient algorithm called IEMiner is designed to discover frequent temporal patterns from interval-based events. The algorithm employs two optimization techniques to reduce the search space and remove non-promising candidates. From the discovered temporal patterns, we build an interval-based classifier called IEClassifier to differentiate closely related classes. Experiments on both synthetic and real world datasets indicate the efficiency and scalability of the proposed approach, as well as the improved accuracy of IEClassifier.

Biological networks such as the protein-protein interaction (PPI) network have been found to contain small recurring subnetworks in significantly higher frequencies than in random networks. Such network motifs are useful for uncovering structural design principles of complex biological networks. However, current network motif finding algorithms models the PPI network as a uni-labeled graph, discovering only unlabeled and thus relatively uninforma-tive network motifs as a result. Our objective is to exploit the currently available biological information that are associated with the vertices (the proteins) to capture not only the topological shapes of the motifs, but also the biological context in which they occurred in the PPI networks for network motif applications. We present a method called LaMoFinder to label network motifs with gene ontology terms in a PPI network. We also show how the resulting labeled network motifs can be used to predict unknown protein functions. Experimental results showed that the labeled network motifs extracted are biologically meaningful and can achieve better performance than existing PPI topology based methods for predicting unknown protein functions.

An automated method for blood vessel segmentation is presented in this paper. The approach uses the nonlinear orthogonal projection to capture the features of vessel networks, and derives a novel local adaptive thresholding algorithm for vessel detection. By embedding in a kind of image decomposition model, the selection of system parameter which reflects the size of concerned convex set is examined. This approach differs from previously known methods in that it uses matched filtering, vessel tracking or supervised methods. The algorithm was tested on two publicly available databases: the DRIVE and the STARE. By comparison with hand-labeled ground truth, good average accuracies are achieved for the both databases.

Cerebral aneurysms are weak or thin spots on blood ves- sels in the brain that balloon out. While the majority of aneurysms do not burst, those that do would lead to se- rious complications including hemorrhagic stroke, perma- nent nerve damage, or death. Yet, surgical options for treat- ing cerebral aneurysms carry high risk to the patient. It is vital for the doctors to accurately diagnose aneurysms that have high probabilities of rupturing. In this applica- tion, the patient dataset has many attributes, ranging from patient profile to results from diagnostic test and features extracted from brain images. Many of the attributes are dis- crete and have missing values. The dataset is also highly biased, with 15% unrupture cases and 85% rupture cases. Building a classifier that unerringly predicts the unrupture (rare) class is a challenge. In this paper, we describe a sys- tematic approach to build such a classifier through suitable combination of data mining algorithms. Our approach au- tomatically determines the optimal combination of these al- gorithms for a dataset. The system has an accuracy of 92% and is currently being deployed at the Huashan Hospital.

Association rule mining aims to discover sets of features that occur together. A variation of association rule mining is ratio rule mining. A ratio rule is an eigenvector of the database that describes ratios of features. However, ratio rules are sensitive to outliers. In this work, we design a prevalence-based model for mining ratio patterns from a database. Our model is more robust to noises, and ratio patterns in our model have clear statistic meanings. We develop an algorithm to quickly determine the sets of features and their ratios that satisfy the prevalence requirement. Data structures, such as hash table and hash tree are utilized to further improve the efficiency of the algorithm. Experiments on synthetic data indicates the efficiency and scalability of the proposed algorithm. We also present a case study on US census data.

Snapshot data have become ubiquitous, e.g., maps, images and videos. By extracting interesting features from snapshot data and analyzing their relative orientations and proximities, we can discover important structure configuration information among groups of features in a snapshot database. In this paper, we introduce a class of pattern called orientation-sensitive patterns, which occur in many applications ranging from weather study, sport game analysis to medical image processing. We examine three approaches to discover orientation-sensitive patterns. We show that the first apriori-based approach is expensive while the second enumeration-based approach is memory intensive. The third approach decomposes an orientation- sensitive pattern into an H-list and a V-list, which greatly simplifies the mining process. Extensive experiment studies show that the third method is more efficient and scalable than the apriori and enumeration algorithms. We also present case studies on soccer game snapshots to demonstrate the interesting patterns discovered.

Estimating the result sizes of XML queries is important in query optimization and is useful in providing a quick feedback about the queries. Existing works have focused on the selectivity estimation of XML queries without order-based axes. In this work, we develop a framework to estimate the result sizes of XPath expressions with order-based axes. We describe how the path and order information of XML elements can be captured and summarized in compact data structures. We also describe methods to estimate the selectivity of XPath queries. The results of extensive experiments on both synthetic and real-world datasets demonstrate the effectiveness and accuracy of the proposed approach.

Existing techniques to mine periodic patterns in time series data are focused on discovering full-cycle periodic patterns from an entire time series. However, many useful partial periodic patterns are hidden in long and complex time series data. In this paper, we aim to discover the partial periodicity in local segments of the time series data. We introduce the notion of character density to partition the time series into variable-length fragments and to determine the lower bound of each character’s period. We propose a novel algorithm, called DPMiner, to find the dense periodic patterns in time series data. Experimental results on both synthetic and real-life datasets demonstrate that the proposed algorithm is effective and efficient to reveal interesting dense periodic patterns.

Existing graph mining algorithms typically assume that databases are relatively static and can fit into the main memory. Mining of subgraphs in a dynamic environment is currently beyond the scope of these algorithms. To bridge this gap, we first introduce a partition-based approach called PartMiner for mining graphs. The PartMiner algorithm finds the frequent subgraphs by dividing the database into smaller and more manageable units, mining frequent subgraphs on these smaller units and finally combining the results of these units to losslessly recover the complete set of subgraphs in the database. Next, we extend PartMiner to handle updates in the dynamic environment. Experimental results indicate that PartMiner is effective and scalable in finding frequent subgraphs, and outperforms existing algorithms in the presence of updates.

The temporal registration of retinal images provides an important groundwork for doctors to monitor the progression of diseases. Retinal image registration is challenging because the intensity of the retina and the vascular structure can vary greatly over time. In this paper, we describe a tree matching approach to register retinal images. We model each vessel in a retinal image as a tree, called vessel feature tree (VFT). We design a matching function to compute the similarity of a pair of vessels based on their VFTs. We develop a global alignment algorithm to compute the best match between the vessels in two images. Experiment results on 300 pairs of real-world retina images indicate that the proposed approach is able to achieve an accuracy of 93%

Recent works in network analysis have revealed the existence of network motifs in biological networks such as the protein-protein interaction (PPI) networks. However, existing motif mining algorithms are not sufficiently scalable to find meso-scale network motifs. Also, there has been little or no work to systematically exploit the extracted network motifs for dissecting the vast interactomes.We describe an efficient network motif discovery algorithm, NeMoFinder, that can mine meso-scale network motifs that are repeated and unique in large PPI networks. Using NeMoFinder, we successfully discovered, for the first time, up to size-12 network motifs in a large whole-genome S. cerevisiae (Yeast) PPI network. We also show that such network motifs can be systematically exploited for indexing the reliability of PPI data that were generated via highly erroneous high-throughput experimental methods.

Many real world objects have states that change over time. By tracking the state sequences of these objects, we can study their behavior and take preventive measures before they reach some undesirable states. In this paper, we propose a new kind of pattern called progressive confident rules to describe sequences of states with an increasing confidence that lead to a particular end state. We give a formal definition of progressive confident rules and their concise set. We devise pruning strategies to reduce the enormous search space. Experiment result shows that the proposed algorithm is efficient and scalable. We also demonstrate the application of progressive confident rules in classification.

A complete set of frequent itemsets can get undesirably large due to redundancy. Several representations have been proposed to eliminate the redundancy. Existing generator based representations rely on a negative border to make the representation lossless. However, negative borders of generators are often very large. The number of itemsets on a negative border sometimes even exceeds the total number of frequent itemsets. In this paper, we propose to use a positive border together with frequent generators to form a lossless representation. A set of frequent generators plus its positive border is always no larger than the corresponding complete set of frequent itemsets, thus it is a true concise representation. The generalized form of this representation is also proposed. We develop an efficient algorithm, called GrGrowth, to mine generators and positive borders as well as their generalizations.

MOTIVATION: Experimental limitations in high-throughput protein-protein interaction detection methods have resulted in low quality interaction datasets that contained sizable fractions of false positives and false negatives. Small-scale, focused experiments are then needed to complement the high-throughput methods to extract true protein interactions. However, the naturally vast interactomes would require much more scalable approaches. RESULTS: We describe a novel method called IRAP* as a computational complement for repurification of the highly erroneous experimentally derived protein interactomes. Our method involves an iterative process of removing interactions that are confidently identified as false positives and adding interactions detected as false negatives into the interactomes. Identification of both false positives and false negatives are performed in IRAP* using interaction confidence measures based on network topological metrics. Potential false positives are identified amongst the detected interactions as those with very low computed confidence values, while potential false negatives are discovered as the undetected interactions with high computed confidence values. Our results from applying IRAP* on large-scale interaction datasets generated by the popular yeast-two-hybrid assays for yeast, fruit fly and worm showed that the computationally repurified interaction datasets contained potentially lower fractions of false positive and false negative errors based on functional homogeneity. AVAILABILITY: The confidence indices for PPIs in yeast, fruit fly and worm as computed by our method can be found at our website http://www.comp.nus.edu.sg/~chenjin/fpfn.

This work addresses the problem of finding boundary points in multidimensional data sets. Boundary points are data points that are located at the margin of densely distributed data such as a cluster. We describe a novel approach called BORDER (a BOundaRy points DEtectoR) to detect such points. BORDER employs the state-of-the-art database technique - the Gorder kNN join and makes use of the special property of the reverse k nearest neighbor (RkNN). Experimental studies on data sets with varying characteristics indicate that BORDER is able to detect the boundary points effectively and efficiently.

Mining topological patterns in spatial databases has received a lot of attention. However, existing work typically ignores the temporal aspect and suffers from certain efficiency problems. They are not scalable for mining topological patterns in spatio-temporal databases. In this paper, we study the problem for mining topological patterns by incorporating the temporal aspect in the mining process. We introduce a summary-structure that records the instances' count information of a feature in a region within a time window. Using this structure, we design an algorithm, TopologyMiner, to find interesting topological patterns without the need to generate candidates. Experimental results show that TopologyMiner is effective and scalable in finding topological patterns and outperforms Apriori-like algorithm by a few orders of magnitudes.

The Reverse k-Nearest Neighbors (RkNN) queries are important in profile-based marketing, information retrieval, decision support and data mining systems. However, they are very expensive and existing algorithms are not scalable to queries in high dimensional spaces or of large values of k. This paper describes an efficient estimation-based RkNN search algorithm (ERkNN) which answers RkNN queries based on local kNN-distance estimation methods. The proposed approach utilizes estimation-based filtering strategy to lower the computation cost of RkNN queries. The results of extensive experiments on both synthetic and real life datasets demonstrate that ERkNN algorithm retrieves RkNN efficiently and is scalable with respect to data dimensionality, k, and data size.

The process of discovering frequent patterns from large semistructured data repositories is one of the hardest categories of tree mining problems, since it involves the discovery of unordered embedded tree patterns. Existing work has focused primarily on the discovery of ordered, induced trees. This work proposes a divide-and-conquer algorithm called WTIMiner to discover the complete set of frequent unordered embedded subtrees. The algorithm successfully reduces the complexity of pattern matching and counting problem that a regular tree mining algorithm faces. Experimental results demonstrate the efficiency and scalability of WTIMiner in terms of both time and space

Efficient evaluation of XML queries requires the determination of whether a relationship exists between two elements. A number of labeling schemes have been designed to label the element nodes such that the relationships between nodes can be easily determined by comparing their labels. With the increased popularity of XML on the Web, finding a labeling scheme that is able to support order-sensitive queries in the presence of dynamic updates becomes urgent. We propose a new labeling scheme that take advantage of the unique property of prime numbers to meet this need. The global order of the nodes can be captured by generating simultaneous congruence values from the prime number node labels. Theoretical analysis of the label size requirements for the various labeling schemes is given. Experiment results indicate that the prime number labeling scheme is compact compared to existing dynamic labeling schemes, and provides efficient support to order-sensitive queries and updates.

Temporal registration of retinal images is helpful to provide physicians important information in tracking the evolution of eye-related diseases. The vascular structure of the retina is the most appropriate feature representation for registration. This paper describes a fast chamfer matching applied to the vascular structure to align pairs of fundus images. While the fast chamfer matching is able to achieve successful alignment consistently, it fails to find correct model parameters in a few cases. To alleviate this problem, we propose a nonparametric elastic matching method. The two matching algorithms are tested on 98 pairs of temporal fundus images. We found that elastic matching gives better performance than the fast chamfer matching method where there are 3 failure cases were reported.

Current protein interaction detection via high-throughput experimental methods such as yeast-two-hybrid has been reported to be highly erroneous. This work introduces a novel measure called IRAP for assessing the reliability of protein interaction based on the underlying topology of the protein interaction network. A candidate protein interaction is considered to be reliable if it is involved in a closed loop in which the alternative path of interactions between the two interacting proteins is strong. We design an algorithm to compute the IRAP value for each interaction in a protein interaction network. Validation of IRAP as a measure for assessing the reliability of protein-protein interactions from conventional high-throughput experiments is performed. We devise a heuristic algorithm to compute IRAP that is able to achieve a 40% speedup in runtime while maintaining a 95% accuracy.

Keyword search is easy to use since it does not require the prior knowledge of query languages or the structure of the underlying data. However, keyword search does not utilize the rich information encoded in the structures of XML to aid in the retrieval of documents. In This work, we devise a context-aware approach for searching XML to improve the effectiveness of keyword search on XML via query expansion. We find a set of XML path expressions that capture the contextual meaning of a keyword query. Paths in the contexts of the query are used to expand the original query to improve the effectiveness of keyword search on XML. Empirical results indicate that the proposed path-augmented keyword search of XML documents outperforms current keyword expansion and keyword proximity search techniques.

XML is increasingly important in data exchange and information management. A large amount of efforts have been spent in developing efficient techniques for storing, querying, indexing and accessing XML documents. In This work we propose a new approach to clustering XML data. In contrast to previous work, which focused on documents defined by different DTDs, the proposed method works for documents with the same DTD. Our approach is to extract features from documents, modeled by ordered labeled trees, and transform the documents to vectors in a high-dimensional Euclidean space based on the occurrences of the features in the documents. We then reduce the dimensionality of the vectors by principal component analysis (PCA) and cluster the vectors in the reduced dimensional space. The PCA enables one to identify vectors with co-occurrent features, thereby enhancing the accuracy of the clustering. Experimental results based on documents obtained from Wisconsin's XML data bank show the effectiveness and good performance of the proposed techniques.

The widespread use of spatio-temporal databases and applications has fuelled an urgent need to discover interesting time and space patterns in such databases. While much work has been done in discovering time/sequence patterns or spatial patterns, discovering of patterns involving both time and space dimensions is still in its infancy, We introduce the concept of flow patterns. Flow patterns are intended to describe the change of events over space and time. These flow patterns are useful to the understanding of many real-life applications. We present a disk-based algorithm, FlowMiner, which utilizes temporal relationships and spatial relationships amid events to generate flow patterns. Our performance study shows that FlowMiner is both scalable and efficient. Experiments on real-life datasets also reveal interesting flow patterns.

Ensemble methods have proved effective to achieve higher accuracy. Some simple ensemble methods, such as Bagging, work well with unstable base algorithms, but fail with stable ones. The reason is that such methods achieve higher accuracy by reducing only the variance of the base algorithms. It does not touch the bias. Here, we propose a novel ensemble method, mode committee, intended to work for both stable and unstable base algorithms. It first derive a new algorithm, called mode competitor, from given base algorithm, with the help of k-modes clustering method and the local learning strategy. Randomness is injected into each mode competitor by the process of random seeding. The aim of deriving mode competitor is to reduce the bias with the possible increasing variance. Then, multiple mode competitors form a committee and vote on the decision of new example, with the aim to reduce the variance of mode competitors. Such an arithmetic framework has been materialized by two base algorithms, the unstable C4.5 and the stable naive Bayes. Extensive empirical results demonstrate this method's superiority, and further analysis by bias-variance decomposition reveals that it is due to the low-bias of mode competitors.

Caching the results of frequent query patterns can improve the performance of query evaluation. This paper describes a 2-pass mining algorithm called 2PXMiner to discover frequent XML query patterns. We design 3 data structures to expedite the mining process. Experiments results indicate that 2PXMiner is both efficient and scalable.

As XML prevails over the Internet, the efficient retrieval of XML data becomes important. Research to improve query response times has been largely concentrate on indexing XML documents and processing regular path expressions. Another approach is to discover frequent query patterns since the answers to these queries can be stored and indexed. Mining frequent query patterns requires more than simple tree matching since the XML queries involves special characters such as "*" or "//". In addition, the matching process can be expensive since the search space is exponential to the size of XML schema. In this paper, we present two mining algorithms, XQPMiner and XQPMinerTID, to discover frequent query pattern frees from a large collection of XML queries efficiently. Both algorithms exploit schema information to guide the enumeration of candidate subtrees, thus eliminating unnecessary node expansions. Experiments results show that the proposed methods are efficient and have good scalability.

Accurate measurement and identification of blood vessels could provide useful information to clinical diagnosis. A piecewise Gaussian model is proposed to describe the intensity distribution of vessel profile in this paper. The characteristic of central reflex is specially considered in the proposed model. The comparison with the single Gaussian model is performed, which shows that the piecewise Gaussian model is a more appropriate model for vessel profile. The obtained model parameters could be utilized in the identification of vessel type. The minimum Mahalanobis distance classifier is employed in the classification. 505 segments of vessels were tested. The success rate is 82.46% and 89.03% for the arteries and veins respectively.

Traditional sequence alignment methods are effective in identifying homologous proteins that are highly similar. However, these approaches do not perform well for remote homologous proteins, that is, proteins whose 3D structures are similar but their sequences are not. Recent biological research reveals that protein sequences contain residues that determine the 3D structure of proteins. In this work, we investigate incorporating this information to aid in the clustering of protein databases. We capture protein residues in the form of patterns with fixed order among them. First, the significant patterns are extracted from the protein sequences. Based on the extracted patterns, we perform sequence mining to generate the order among them. Finally, we adopt a partition-based method to cluster protein sequences using the patterns and order features. Experiments on COG and SCOP40 datasets show that our new approach is able to generate high quality clusters that are similar to those determined manually by the biologists.

Traditional classification rules are in the form of production rules. Recent works in hybrid classification algorithms have proposed the generation of contextual rules, whereby the right-hand side of the production rule is replaced by a classifier, to achieve higher accuracy. In this work, we present a framework to further generalize classification rules such that the left-hand side of a production rule is expressed as a conjunction of classifiers, called space splitters. An intelligent divide-and-conquer approach is designed to construct such generalized classification rules. The construction algorithm, GCTree, is elegant, efficient and scalable. The resulting classifier is able to achieve high predictive accuracy that outperforms naive Bayes and C4.5. Experiments demonstrate that GCTree is compact and stable.

Spatial data mining is the extraction of implicit knowledge, spatial relations or other patterns not explicitly stored in spatial database. The focus of this paper is placed on the information derivation of spatial data. Geographical coordinates of hot spots in forest fire regions, which are extracted from the satellite images, are studied and used in the detection of likely fire points. False alarms can occur in the derived hotspots. While this false information can be identified by comparing the radiance detected at several bands, we introduce a different approach to remove some of the false alarms. We use clustering and a Hough transformation to determine regular patterns in the derived hotspots and classify them as false alarms on the assumption that fires usually do not spread in regular patterns such as in a straight line. This project demonstrates the application of spatial data mining to reduce false alarms from the set of hotspots derived from NOAA images.

The increasing number of image repositories has made image mining an important task because of its potential in discovering useful image patterns from a large set of images. In this paper, we introduce the notion of viewpoint patterns for image databases. Viewpoint patterns refer to patterns that capture the invariant relationships of one object from the point of view of another object. These patterns are unique and significant in images because the absolute positional information of objects for most images is not important, but rather, it is the relative distance and orientation of the objects from each other that is meaningful. We design a scalable and efficient algorithm to discover such viewpoint patterns. Experiments results on various image sets demonstrate that viewpoint patterns are meaningful and interesting to human users.

MOTIVATION: The function of an unknown biological sequence can often be accurately inferred if we are able to map this unknown sequence to its corresponding homologous family. At present, discriminative methods such as SVM-Fisher and SVM-pairwise, which combine support vector machine (SVM) and sequence similarity, are recognized as the most accurate methods, with SVM-pairwise being the most accurate. However, these methods typically encode sequence information into their feature vectors and ignore the structure information. They are also computationally inefficient. Based on these observations, we present an alternative method for SVM-based protein classification. Our proposed method, SVM-I-sites, utilizes structure similarity for remote homology detection. RESULT: We run experiments on the Structural Classification of Proteins 1.53 data set. The results show that SVM-I-sites is more efficient than SVM-pairwise. Further, we find that SVM-I-sites outperforms sequence-based methods such as PSI-BLAST, SAM, and SVM-Fisher while achieving a comparable performance with SVM-pairwise. AVAILABILITY: I-sites server is accessible through the web at http://www.bioinfo.rpi.edu. Programs are available upon request for academics. Licensing agreements are available for commercial interests. The framework of encoding local structure into feature vector is available upon request.

Traditionally, XML documents are processed at where they are stored. This allows the query processor to exploit pre-computed data structures (e.g., index) to retrieve the desired data efficiently. However, this mode of processing is not suitable for many applications where the documents are frequently updated. In such situations, efficient evaluation of multiple queries over streaming XML documents becomes important. This paper introduces a new operator, mqX-scan, which efficiently evaluates multiple queries with a single pass on streaming XML data. To facilitate matching, mqX-scan utilizes templates containing paths that have been traversed to match regular path expression patterns in a pool of queries. Results of the experiments demonstrate the efficiency and scalability of the mqX-scan operator.

It is increasingly important to develop scalable integration techniques for the growing number of XML data sources. A practical starting point for the integration of large numbers of Document Type Definitions (DTDs) of XML sources would be to first find clusters of DTDs that are similar in structure and semantics. Reconciling similar DTDs within such a cluster will be an easier task than reconciling DTDs that are different in structure and semantics as the latter would involve more restructuring. We introduce XClust, a novel integration strategy that involves the clustering of DTDs. A matching algorithm based on the semantics, immediate descendents and leaf-context similarity of DTD elements is developed. Our experiments to integrate real world DTDs demonstrate the effectiveness of the XClust approach.

Advances in imaging techniques have led to large repositories of images. There is an increasing demand for automated systems that can analyze complex medical images and extract meaningful information for mining patterns. Here, we describe a real-life image mining application to the problem of tumour cell counting. The quantitative analysis of tumour cells is fundamental to characterizing the activity of tumour cells. Existing approaches are mostly manual, time-consuming and subjective. Efforts to automate the process of cell counting have largely focused on using image processing techniques only. Our studies indicate that image processing alone is unable to give accurate results. In this paper, we examine the use of extracted features rules to aid in the process of tumor cell counting. We propose a robust local adaptive thresholding and dynamic water immersion algorithms to segment regions of interesting from background. Meaningful features are then extracted from the segmented regions. A number of base classifiers are built to generate features rules to help identify the tumor cell. Two voting strategies are implemented to combine the base classifiers into a meta-classifier. Experiment results indicate that this process of using extracted features rules to help identify tumor cell leads to better accuracy than pure image processing techniques alone.

Concept lattice model, the core structure in formal concept analysis, has been successfully applied in software engineering and knowledge discovery. This paper integrates the simple base classifier (Naïve Bayes or Nearest Neighbour) into each node of the concept lattice to form a new composite classifier. Two new classification systems are developed, CLNB and CLNN, which employ efficient constraints to search for interesting patterns and voting strategy to classify a new object. CLNB integrates the Naïïve Bayes base classifier into concept nodes while CLNN incorporates the Nearest Neighbour base classifier into concept nodes. Experimental results indicate that these two composite classifiers greatly improve the accuracy of their corresponding base classifier. In addition, CLNB even outperforms three other state-of-the-art classification methods, NBTree, CBA and C4.5 Rules.

Diabetic retinopathy is a major cause of blindness in the world. Regular screening and timely intervention can halt or reverse the progression of this disease. Digital retinal imaging technologies have become an integral part of eye screening programs worldwide due to their greater accuracy and repeatability in staging diabetic retinopathy. These screening programs produce an enormous number of retinal images since diabetic patients typically have both their eyes examined at least once a year. Automated detection of retinal lesions can reduce the workload and increase the efficiency of doctors and other eye-care personnel reading the retinal images and facilitate the follow-up management of diabetic patients. Existing techniques to detect retinal lesions are neither adaptable nor sufficiently sensitive and specific for real-life screening application. In this paper, we demonstrate the role of domain knowledge in improving the accuracy and robustness of detection of hard exudates in retinal images. Experiments on 543 consecutive retinal images of diabetic patients indicate that we are able to achieve 100% sensitivity and 74% specificity in the detection of hard exudates.

Building predictive models and finding useful rules are two important tasks of data mining. While building predictive models has been well studied, finding useful rules for action still presents a major problem. A main obstacle is that many data mining algorithms often produce too many rules. Existing research has shown that most of the discovered rules are actually redundant or insignificant. Pruning techniques have been developed to remove those spurious and/or insignificant rules. In this paper, we argue that being a significant rule (or a non-redundant rule), however, does not mean that it is a potentially useful rule for action. Many significant rules (unpruned rules) are in fact not actionable. This paper studies this issue and presents an efficient algorithm to identify these non-actionable rules. Experiment results on many real-life datasets show that the number of non-actionable rules is typically quite large. The proposed technique thus enables the user to focus on fewer rules and to be assured that the remaining rules are non-redundant and potentially useful for action.

The world around us changes constantly. Knowing what has changed is an important part of our lives. For businesses, recognizing changes is also crucial. It allows businesses to adapt themselves to the changing market needs. In this paper, we study changes of association rules from one time period to another. One approach is to compare the supports and/or confidences of each rule in the two time periods and report the differences. This technique, however, is too simplistic as it tends to report a huge number of rule changes, and many of them are, in fact, simply the snowball effect of a small subset of fundamental changes. Here, we present a technique to highlight the small subset of fundamental changes. A change is fundamental if it cannot be explained by some other changes. The proposed technique has been applied to a number of real-life datasets. Experiments results show that the number of rules whose changes are unexplainable is quite small (about 20% of the total number of changes discovered), and many of these unexplainable changes reflect some fundamental shifts in the application domain.

Diabetic-related eye diseases are the most common cause of blindness in the world. So far the most effective treatment for these eye diseases is early detection through regular screening. To lower the cost of such screenings, we employ state-of-the-art image processing techniques to automatically detect the presence of abnormalities in the retinal images obtained during the screenings. The authors focus on one of the abnormal signs: the presence of exudates/lesions in the retinal images. We propose a novel approach that combines brightness adjustment procedure with statistical classification method and local-window-based verification strategy. Experimental results indicate that we are able to achieve 100% accuracy in terms of identifying all the retinal images with exudates while maintaining a 70% accuracy in correctly classifying the truly normal retinal images as normal. This translates to a huge amount of savings in terms of the number of retinal images that need to be manually reviewed by the medical professionals each year.

Real-life data mining applications are interesting because they often present a different set of problems for data miners.One such real-life application that we have done is on the diabetic patients databases.Valuable lessons are learnt from this application.In particular, we discover that the often neglected pre-processing and post-processing steps in knowledge discovery are the most critical elements in determining the success of a real-life data mining application.In this paper, we shall discuss how we carry out knowledge discovery on this diabetic patient database, the interesting issues that have surfaced, as well as the lessons we have learnt from this application.We will describe a semi-automatic means for cleaning the diabetic patient database, and present a step-by-step approach to help the health doctors explore their data and to understand the discovered rules better.While it is important to generate understandable rules, it is also important to the medical doctors to have a complete picture of all

Many existing data mining techniques often produce a large number of rules, which make it very difficult for manual inspection of the rules to identify those interesting ones. This problem represents a major gap between the results of data mining and the understanding and use of the mining results. In this paper, we argue that the key problem is not with the large number of rules because if there are indeed many rules that exist in data, they should be discovered. The main problem is with our inability to organize, summarize and present the rules in such a way that they can be easily analyzed by the user. In this paper, we propose a technique to intuitively organize and summarize the discovered rules. With this organization, the discovered rules can be presented to the user in the way as we think and talk about knowledge in our daily lives. This organization also allows the user to view the discovered rules at different levels of details, and to focus his/her attention on those interes...

There is an increasing demand for systems that can automatically analyze images and extract semantically meaningful information. IRIS, an Integrated Retinal Information system, has been developed to provide medical professionals easy and unified access to the screening, trend and progression of diabetic-related eye diseases in a diabetic patient database. This paper shows how mining techniques can be used to accurately extract features in the retinal images. In particular, we apply a classification approach to determine the conditions for tortuousity in retinal blood vessels.

Association rules are a fundamental class of patterns that exist in data. The key strength of association rule mining is its completeness. It finds all associations in the data that satisfy the user specified minimum support and minimum confidence constraints. This strength, however, comes with a major drawback. It often produces a huge number of associations. This is particularly true for data sets whose attributes are highly correlated. The huge number of associations makes it very difficult, if not impossible, for a human user to analyze in order to identify those interesting/useful ones. In this paper, we propose a novel technique to overcome this problem. The technique first prunes the discovered associations to remove those insignificant associations, and then finds a special subset of the unpruned associations to form a summary of the discovered associations. We call this subset of associations the direction setting (DS) rules as they set the directions that are followed by the...

Association rule mining is an important model in data mining. Its mining algorithms discover all item associations (or rules) in the data that satisfy the user-specified minimum support (minsup) and minimum confidence (minconf) constraints. Minsup controls the minimum number of data cases that a rule must cover. Minconf controls the predictive strength of the rule. Since only one minsup is used for the whole database, the model implicitly assumes that all items in the data are of the same nature and/or have similar frequencies in the data. This is, however, seldom the case in reallife applications. In many applications, some items appear very frequently in the data, while others rarely appear. If minsup is set too high, those rules that involve rare items will not be found. To find rules that involve both frequent and rare items, minsup has to be set very low. This may cause combinatorial explosion because those frequent items will be associated with one another in all possible ways. T...

1. Introduction Data mining aims to develop a new generation of tools tointelligently assist humans in analyzing mountains of data.

One of the major problems in the field of knowledge discovery (or data mining) is the interestingness problem. Past research and applications have found that, in practice, it is all too easy to discover a huge number of patterns in a database. Most of these patterns are actually useless or uninteresting to the user. But due to the huge number of patterns, it is difficult for the user to comprehend them and to identify those interesting to him/her. To prevent the user from being overwhelmed by the large number of patterns, techniques are needed to rank them according to their interestingness. In this paper, we propose such a technique, called the user-expectation method. In this technique, the user is first asked to provide his/her expected patterns according to his/her past knowledge or intuitive feelings. Given these expectations, the system uses a fuzzy matching technique to match the discovered patterns against the user's expectations, and then rank the discovered patterns according to the matching results. A variety of rankings can be performed for different purposes, such as to confirm the user's knowledge and to identify unexpected patterns, which are by definition interesting. The proposed technique is general and interactive.

In this paper, KPN (Knowledge Petri Net), a unified Petri net model for atemporal and temporal knowledge representation and reasoning, is presented. In the model, logical, causal, uncertain/imprecise, and temporal knowledge are represented in a coherent way. Based on T-invariant computation (a classical linear algebra analysis technique of Petri nets), an efficient and simple algorithm for reasoning on KPN nets is also proposed. The model is the first attempt to handle both atemporal and temporal knowledge within a Petri net framework.

Traditionally, design for assembly is done during the detailed design phase. A designer first maps a set of design requirements into a set of components or subassemblies that can satisfy the given set of requirements. The components and subassemblies are then examined individually to determine whether they conform to the principles of design for assembly. Usually, local changes are performed so that the resultant components/subassemblies are better for assembly. In this paper, we propose to bring the design for assembly analysis into an even earlier phase-that of the conceptual design phase. We argue that by incorporating the design for assembly analysis at the conceptual design phase, we can achieve a more substantial savings as compared to the savings obtained when the design for assembly analysis is only performed as late as the detailed design phase. The basic idea is to select a combination of design concepts (previously stored in a library) such that together they can achieve the stated functional requirements (in the form of state transitional graph) at the minimum cost for assembly. This problem of selecting the right combination of design concepts is reduced to the well-known set covering problem. With this reduction, many existing graph algorithms can be applied to aid in the design for assembly analysis.

This paper presents a systematic procedure for generating the goal region of a mating action from a model of assemblies. The goal region of a mating action is defined as the acceptable destination of the moving object and is used to identify the successful situation of the mating action. In this paper, goal regions are constructed by using the mating features and the constraint types residing in the model of assemblies. The mating features identify the important variables between which the interdependencies are considered, and the constraint types identify the necessary constraints that mating features cannot provide. An analytical solution for C-space interior is developed to construct the goal region from 2D features. The effects of interference and fine-motion planning on success probabilities are also taken into consideration to define theoretical goal regions; goal regions are also subject to uncertainties. An approach that shrinks the nominal boundary of a goal region is also proposed to compute the expectation of success probabilities.

With the increased competition in today's world market, emphasis has been on the ability to shift from an existing paradigm to a new paradigm so as to create new opportunities and to gain new market. A successful paradigm shift is dependent on two factors: (1) the ability to pinpoint the inherent weaknesses in the existing paradigm, (2) the ability to find a paradigm that can replace the old paradigm. In this paper, we show how the integrated feedback approach is able to address these two concerns. The integrated feedback approach was proposed to integrate the design phase with the downstream activities so as to achieve a design that is better for assembly. The approach operates in two phases: an evaluation phase and a redesign suggestion generation phase. A number of objective criteria have been proposed to evaluate a given design from the functional perspective, from the assembly plan perspective, and from the tolerance perspective. Through the evaluation process, weaknesses in the design are identified and techniques for generating feasible redesign suggestions are examined. By encouraging greater participation from the designer, a systematic aid to paradigm shifting can be obtained. A prototype system implementing the integrated feedback approach has been developed on a Sun Sparcstation with graphics simulation on a Silicon Graphics Iris workstation. A real life product, the telephone, is used as an illustration.

Proceedings of the ACM International Multimedia Conference & Exhibition

The authors examine issues involved in integrating the design level and the assembly planning level with a feedback loop. The integration is performed in two stages. The first stage focuses on the evaluation of an assembly plan. Evaluation criteria that can pinpoint areas which need redesign are defined. The second stage is to use the evaluation results to come up with the actual redesign. Algorithms are developed for performing evaluation of assembly plans. From the evaluation results, means of generating hints for redesign are discussed. The hints are then processed and calls are made to the redesign operators to perform the actual redesign of components. The integrated design-planning system has the ability to identify parts that need redesign and the ability to come up with feasible redesign options in polynomial time.>