Mem-PAL: Towards Memory-based Personalized Dialogue Assistants for Long-term User-Agent Interaction

Evaluating the quality of responses generated by open-domain conversation systems is a challenging task. This is partly because there can be multiple appropriate responses to a given dialogue history. Reference-based metrics that rely on comparisons to a set of known correct responses often fail to account for this variety, and consequently correlate poorly with human judgment. To address this problem, researchers have investigated the possibility of assessing response quality without using a set of known correct responses. RUBER demonstrated that an automatic response evaluation model could be made using unsupervised learning for the next-utterance prediction (NUP) task. For the unsupervised learning of such model, we propose a method of manipulating a golden response to create a new negative response that is designed to be inappropriate within the context while maintaining high similarity with the original golden response. We find, from our experiments on English datasets, that using the negative samples generated by our method alongside random negative samples can increase the model’s correlation with human evaluations. The process of generating such negative samples is automated and does not rely on human annotation.

The edge computing works as an interface between the real world data, sensor nodes and cloud devices and provides intended security and privacy features. Use of smart wearable devices embedded with modern chipsets and leveraged with edge computing provides near real time monitoring, display diagnostics of state of the patient health and prevents spread of disease, even if the doctor is distantly located. In this paper, a secure smart health care system is proposed that is supported by multiple physiological sensors, smart devices and edge nodes at patient and doctor end to collect, protect, process, analyze and leverages cloud facilities to manage health records of patients. The analyzed data and requisites data are made available to the patient's smart devices. The simulation results show that the proposed system demands low latency for patients under different medical condition to enable fast and real-time diagnosis from healthcare professionals.

In this paper, we explore the problem of developing personalized chatbots. A personalized chatbot is designed as a digital chatting assistant for a user. The key characteristic of a personalized chatbot is that it should have a consistent personality with the corresponding user. It can talk the same way as the user when it is delegated to respond to others' messages. We present a retrieval-based personalized chatbot model, namely IMPChat, to learn an implicit user profile from the user's dialogue history. We argue that the implicit user profile is superior to the explicit user profile regarding accessibility and flexibility. IMPChat aims to learn an implicit user profile through modeling user's personalized language style and personalized preferences separately. To learn a user's personalized language style, we elaborately build language models from shallow to deep using the user's historical responses; To model a user's personalized preferences, we explore the conditional relations underneath each post-response pair of the user. The personalized preferences are dynamic and context-aware: we assign higher weights to those historical pairs that are topically related to the current query when aggregating the personalized preferences. We match each response candidate with the personalized language style and personalized preference, respectively, and fuse the two matching signals to determine the final ranking score. Comprehensive experiments on two large datasets show that our method outperforms all baseline models.

Two-dimensional (2-D) autoregressive moving average (ARMA) models are commonly applied to describe real-world image data, usually assuming Gaussian or symmetric noise. However, real-world data often present non-Gaussian signals, with asymmetrical distributions and strictly positive values. In particular, SAR images are known to be well characterized by the Rayleigh distribution. In this context, the ARMA model tailored for 2-D Rayleigh-distributed data is introduced—the 2-D RARMA model. The 2-D RARMA model is derived and conditional likelihood inferences are discussed. The proposed model was submitted to extensive Monte Carlo simulations to evaluate the performance of the conditional maximum likelihood estimators. Moreover, in the context of SAR image processing, two comprehensive numerical experiments were performed comparing anomaly detection and image modeling results of the proposed model with traditional 2-D ARMA models and competing methods in the literature.

Unsupervised machine learning approaches hold great promise for recognizing dialogue acts, but the performance of these models tends to be much lower than the accuracies reached by supervised models. However, some dialogues, such as task-oriented dialogues with parallel task streams, hold rich information that has not yet been leveraged within unsupervised dialogue act models. This paper investigates incorporating task features into an unsupervised dialogue act model trained on a corpus of human tutoring in introductory computer science. Experimental results show that incorporating task features and dialogue history features significantly improve unsupervised dialogue act classification, particularly within a hierarchical framework that gives prominence to dialogue history. This work constitutes a step toward building high-performing unsupervised dialogue act models that will be used in the next generation of task-oriented dialogue systems.

Self-supervised learning based transformer and convolution hybrid network for one-shot organ segmentation

One-shot organ segmentation (OS2) aims at segmenting the desired organ regions from the input medi-cal imaging data with only one pre-annotated example as the reference. By using the minimal annotation data to facilitate organ segmentation, OS2 receives great attention in the medical image analysis community due to its weak re-quirement on human annotation. In OS2, one core issue is to explore the mutual information between the support (ref-erence slice) and the query (test slice). Existing methods rely heavily on the similarity between slices, and additional slice allocation mechanisms need to be designed to reduce the impact of the similarity between slices on the segmen-tation performance. To address this issue, we build a novel support-query interactive embedding (SQIE) module, which is equipped with the channel-wise co-attention, spatial-wise co-attention, and spatial bias transformation blocks to identify "what to look", "where to look", and "how to look" in the input test slice. By combining the three mechanisms, we can mine the interactive information of the intersection area and the disputed area between slices, and establish the feature connection between the target in slices with low similarity. We also propose a self-supervised contrastive learning framework, which transforms knowledge from the physical position to the embedding space to facilitate the self-supervised interactive embedding of the query and support slices. Comprehensive experiments on two large benchmarks demonstrate the superior capacity of the pro-posed approach when compared with the current alterna-tives and baseline models.

Affect-aware socially assistive robotics (SAR) has shown great potential for augmenting interventions for children with autism spectrum disorders (ASD). However, current SAR cannot yet perceive the unique and diverse set of atypical cognitive-affective behaviors from children with ASD in an automatic and personalized fashion in long-term (multi-session) real-world interactions. To bridge this gap, this work designed and validated personalized models of arousal and valence for children with ASD using a multi-session in-home dataset of SAR interventions. By training machine learning (ML) algorithms with supervised domain adaptation (s-DA), the personalized models were able to tradeoff between the limited individual data and the more abundant less personal data pooled from other study participants. We evaluated the effects of personalization on a long-term multimodal dataset consisting of four children with ASD with a total of 19 sessions, and derived inter-rater reliability (IR) scores for binary arousal (IR = 83%) and valence (IR = 81%) labels between human annotators. Our results show that personalized Gradient Boosted Decision Trees (XGBoost) models with s-DA outperformed two non-personalized individualized and generic model baselines not only on the weighted average of all sessions, but also statistically ( p < .05) across individual sessions. This work paves the way for the development of personalized autonomous SAR systems tailored toward individuals with atypical cognitive-affective and socio-emotional needs.

Breast cancer recurrence prediction remains a critical challenge in oncological surveillance and personalized treatment planning. Accurate temporal modeling of recurrence risk requires sophisticated handling of longitudinal patient data characterized by irregular time intervals, missing values, and complex temporal dependencies. This study presents an attention-enhanced Long Short-Term Memory (LSTM) framework optimized for breast cancer recurrence time prediction. The proposed architecture integrates multi-head self-attention mechanisms with bidirectional LSTM networks to capture critical prognostic temporal patterns from electronic health records. We incorporate time-aware position encoding and specialized loss functions combining survival analysis metrics with cross-entropy objectives. Comprehensive experiments on longitudinal breast cancer datasets demonstrate superior predictive performance compared to traditional Cox proportional hazards models and standard recurrent neural networks. The attention mechanism successfully identifies key temporal biomarker changes and clinical events contributing to recurrence risk, achieving an overall C-index of 0.891. Ablation studies confirm the substantial contribution of attention-based temporal modeling, with improvements of 7.3% in time-dependent AUC over baseline LSTM architectures. The interpretable attention weights provide clinically actionable insights for oncologists in developing personalized surveillance strategies. This research advances temporal deep learning methodologies for cancer prognosis and establishes a foundation for AI-driven recurrence monitoring systems in clinical practice.

Dependency-based anomaly detection methods detect anomalies by looking at the deviations from the normal probabilistic dependency among variables and are able to discover more subtle and meaningful anomalies. However, with high dimensional data, they face two key challenges. One is how to find the right set of relevant variables for a given variable from the large search space to assess dependency deviation. The other is how to use the dependency to estimate the expected value of a variable accurately. In this paper, we propose the Local Prediction approach to Anomaly Detection (LoPAD) framework to deal with the two challenges simultaneously. Through introducing Markov Blanket into dependency-based anomaly detection, LoPAD decomposes the high dimensional unsupervised anomaly detection problem into local feature selection and prediction problems while achieving better performance and interpretability. The framework enables instantiations with off-the-shelf predictive models for anomaly detection. Comprehensive experiments have been done on both synthetic and real-world data. The results show that LoPAD outperforms state-of-the-art anomaly detection methods.

A general soft-balanced clustering framework based on a novel balance regularizer

Scene-text image synthesis techniques that aim to naturally compose text instances on background scene images are very appealing for training deep neural networks due to their ability to provide accurate and comprehensive annotation information. Prior studies have explored generating synthetic text images on two-dimensional and three-dimensional surfaces using rules derived from real-world observations. Some of these studies have proposed generating scene-text images through learning; however, owing to the absence of a suitable training dataset, unsupervised frameworks have been explored to learn from existing real-world data, which might not yield reliable performance. To ease this dilemma and facilitate research on learning-based scene text synthesis, we introduce DecompST, a real-world dataset prepared from some public benchmarks, containing three types of annotations: quadrilateral-level BBoxes, stroke-level text masks, and text-erased images. Leveraging the DecompST dataset, we propose a Learning-Based Text Synthesis engine (LBTS) that includes a text location proposal network (TLPNet) and a text appearance adaptation network (TAANet). TLPNet first predicts the suitable regions for text embedding, after which TAANet adaptively adjusts the geometry and color of the text instance to match the background context. After training, those networks can be integrated and utilized to generate the synthetic dataset for scene text analysis tasks. Comprehensive experiments were conducted to validate the effectiveness of the proposed LBTS along with existing methods, and the experimental results indicate the proposed LBTS can generate better pretraining data for scene text detectors. Our dataset and code are made available at: https://github.com/iiclab/DecompST.

Accurate customer LifeTime Value (LTV) predictions are crucial for customer relationship management, especially in Supply Chain Platforms (SCP), which involve effectively managing the service resources in business decision-making. Previous LTV prediction methods usually rely on ample historical customer data, which is not available in the early stages of a customer's lifecycle. It makes the modeling of the historical customer data a difficult task due to the data sparsity. Besides, the long-tail distribution of customer LTV also brings new challenges to the prediction of LTV. To tackle the above issues, we propose CDLtvS, a novel Cross Domain method for customer Lifetime value prediction in SCP. It leverages rich cross-domain information from upstream platforms to enhance LTV predictions in downstream platforms. Firstly, CDLtvS pre-trains the customer representations by an LTV modeling framework named LtvS in source and target domains separately. Specifically, LtvS incorporates the Expert Mask Network (ExMN), which not only effectively models the long-tail distribution of LTV in single-domain but also resolves cross-domain learning model bias resulting from this distribution. Then, the various-level alignment mechanism is introduced to keep the consistency of knowledge transferring from source to target domains on both sparse and non-sparse data. Comprehensive experiments on real-world data from JD, one of the world's largest supply chain platforms, demonstrate that CDLtvS achieves a normalized mean average error of 0.3378 in LTV prediction, outperforming 16.3% to the baseline. Additionally, the improvements of ≥2.3% across various data sparsity levels (0% -- 80%) provide valuable insights into cross-domain LTV modeling.

Recent years have witnessed the rapidly growing of the amount of semistructured documents in real-world applications. Due to the huge size of the real-world data, how to manage semistructured documents effectively is a big challenge for researchers. As a fundamental task in natural language processing field, document classification is a feasible way to handle the large-scale semistructured documents. However, existing methods fail to explicitly take advantage of the hierarchical semantics in semistructured documents. It's known that the contained semantics is beneficial for understanding the semistructured documents. Considering the hierarchical structure of a given semistructured document, we propose a semistructured document classification framework which explicitly utilizes the semantic hierarchical attention mechanism. More specifically, the hierarchical attention mechanism and graph neural network are employed to model semistructured documents, by which the multilevel semantic relationships and grammatical information are considered. Moreover, we propose an adaptive class cost learning method to treat the issue of data imbalance. Comprehensive experiments are conducted on two real-world data sets, and the results demonstrate that our framework performs better than selected baselines for semistructured document classification.

Commodity ultra-high-frequency (UHF) RFID authentication systems only provide weak user authentication, as RFID tags can be easily stolen, lost, or cloned by attackers. This paper presents the design and evaluation of SmartRFID, a novel UHF RFID authentication system to promote commodity crypto-less UHF RFID tags for security-sensitive applications. SmartRFID explores extremely popular smart devices and requires a legitimate user to enroll his smart device along with his RFID tag. Besides authenticating the RFID tag as usual, SmartRFID verifies whether the user simultaneously possesses the associated smart device with both feature-based machine learning and deep learning techniques. The user is considered authentic if and only if passing the dual verifications. Comprehensive user experiments on commodity smartwatches and RFID devices confirmed the high security and usability of SmartRFID. In particular, SmartRFID achieves a true acceptance rate of above 97.5% and a false acceptance rate of less than 0.7% based on deep learning. In addition, SmartRFID can achieve an average authentication latency of less than 2.21 s, which is comparable to inputting a PIN on a door keypad or smartphone.