Quanyu Long lists you as their PhD advisor. Claim this profile to confirm it and keep the rest of your record right.
Claim this profileAcademic lineage
View as a treeStudents and postdocs3
Possible advisorsa guess from early papers, not confirmed
- Qiang YangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works180 from public data
- 24,193
- 4,891
- 1,544
- Cross-domain sentiment classification via spectral feature alignment837
This work develops a general solution to sentiment classification when the authors do not have any labels in a target domain but have some labeled data in a different domain, regarded as source domain and proposes a spectral feature alignment (SFA) algorithm to align domain-specific words from different domains into unified clusters, with the help of domain-independent words as a bridge.
- Transfer learning via dimensionality reduction727
A new dimensionality reduction method is proposed to find a latent space, which minimizes the distance between distributions of the data in different domains in a latentspace, which can be treated as a bridge of transferring knowledge from the source domain to the target domain.
- 631
- 554
- Coupled Multi-Layer Attentions for Co-Extraction of Aspect and Opinion Terms448
A novel deep learning model, named coupled multi-layer attentions, where each layer consists of a couple of attentions with tensor operators that are learned interactively to dually propagate information between aspect terms and opinion terms.
- Recursive Neural Conditional Random Fields for Aspect-based Sentiment Analysis431
A novel joint model that integrates recursive neural networks and conditional random fields into a unified framework for explicit aspect and opinion terms co-extraction and is flexible to incorporate hand-crafted features into the proposed model to further boost its information extraction performance.
- AdaRNN389
This paper proposes Adaptive RNNs (AdaRNN) to tackle the TCS problem by building an adaptive model that generalizes well on the unseen test data and proposes Temporal Distribution Characterization to better characterize the distribution information in the TS.
- Heterogeneous Transfer Learning for Image Classification385
This paper proposes a heterogeneous transfer learning framework for knowledge transfer between text and images by enriching the representation of the target images with semantic concepts extracted from the auxiliary source data through a novel matrix factorization method.
- Supervised Representation Learning: Transfer Learning with Deep Autoencoders372
This paper proposes a supervised representation learning method based on deep autoencoders for transfer learning that consists of an embedding layer and a label encoding layer that minimize the difference between domains explicitly and encode label information in learning the representation.
- 312
- HYDRA: Massively Compositional Model for Cross-Project Defect Prediction295
Improvements of HYDRA over other baseline approaches in terms of F1-score and when inspecting the top 20 percent lines of code are substantial, and in most cases the improvements are significant and have large effect sizes across the 29 datasets.
- Multilingual Jailbreak Challenges in Large Language Models283
Experimental results show that ChatGPT fine-tuned with such data can achieve a substantial reduction in unsafe content generation, and a novel self-defense framework that automatically generates multilingual training data for safety fine-tuning is proposed.
- 253
- 214
- Cross-Domain Co-Extraction of Sentiment and Topic Lexicons213
Experimental results show that the proposed domain adaptation framework for sentiment- and topic- lexicon co-extraction can extract precise lexicons in the target domain without any annotation.
- Distant Domain Transfer Learning203
Inspired by the cognitive process of human where two seemingly unrelated concepts can be connected by learning intermediate concepts gradually, a Selective Learning Algorithm (SLA) is proposed to solve the DDTL problem with supervised autoencoder or supervised convolutional autoen coder as a base model for handling different types of inputs.
- Adaptive Transfer Learning202
An Adaptive Transfer learning algorithm based on Gaussian Processes (AT-GP), which can be used to adapt the transfer learning schemes by automatically estimating the similarity between a source and a target task, is proposed.
- Short and sparse text topic modeling via self-aggregation201
A novel model integrating topic modeling with short text aggregation during topic inference is presented, founded on general topical affinity of texts rather than particular heuristics, making the model readily applicable to various short texts.
- Hybrid Heterogeneous Transfer Learning through Deep Learning201
A deep learning approach to learn a feature mapping between cross-domain heterogeneous features as well as a better feature representation for mapped data to reduce the bias issue caused by the cross- domain correspondences.
- Heterogeneous Domain Adaptation for Multiple Classes161
An efficient multi-class heterogeneous domain adaptation method, where data from source and target domains are represented by heterogeneous features of different dimensions, to reconstruct a sparse feature transformation matrix to map the weight vector of classifiers learned from the source domain to the target domain.
- Multi-task Learning145
A similarity measure is developed that can quantify transference among tasks and use this quantity to both better understand the optimization dynamics of multi-task learning as well as improve overall learning performance.
- 137
- Adaptive localization in a dynamic WiFi environment through multi-view learning133
LeManCoR, a system for adapting the mapping function between the signal space and physical location space over different time periods based on Manifold Co-Regularization is described and it is shown that LeMan co-CoR can effectively transfer the knowledge between two time periods without requiring too much new calibration effort.
- Knowledge Transfer for Deep Reinforcement Learning with Hierarchical Experience Replay127
A new policy distillation architecture for deep reinforcement learning is proposed, where it is assumed that each task uses its task-specific high-level convolutional features as the inputs to the multi-task policy network.
- Transfer learning for WiFi-based indoor localization123
It is shown that WILP can be modeled as a transfer learning problem for regression modeling, where several important cases of knowledge transfer can be identified that range from transferring the localization models over time, across space and across client devices.
- Estimating Location Using Wi-Fi119
This department presents the results of the first Data Mining Contest held at the 2007 International Conference on Data Mining.
- Active Transfer Learning for Cross-System Recommendation114
This paper proposes a framework to construct entity correspondence with limited budget by using active learning to facilitate knowledge transfer across recommender systems, and first iteratively select entities in the target system based on a proposed criterion to query their correspondences in the source system.
- Exploiting Social Relations and Sentiment for Stock Prediction113
Results demonstrate that topic senti- ments from close neighbors are able to help prove the prediction of a stock markedly and show that social sentiment about stock topics and stock relationship topics are predictive of each stock's market.
- Transferring multi-device localization models using latent multi-task learning109
A latent multi-task learning algorithm that treats multiple devices as multiple learning tasks, and requires the hypotheses learned in a latent feature space are similar to solve the multi-device indoor localization problem.
- Latent Independent Excitation for Generalizable Sensor-based Cross-Person Activity Recognition107
A novel method named Generalizable Independent Latent Excitation (GILE) for human activity recognition is introduced, which greatly enhances the cross-person generalization capability of the model and is superior to existing methods in the sense that it does not require any access to the target domain information.
- 92
- Real world activity recognition with multiple goals89
This article asks the fundamental questions of whether users often carry out multiple concurrent and interleaving activities or single activities in their daily life, and if so, whether such complex behavior can be detected accurately using sensors.
- 88
- 88
- Recursive Neural Structural Correspondence Network for Cross-domain Aspect and Opinion Co-Extraction88
A novel recursive neural network is developed that could reduce domain shift effectively in word level through syntactic relations across domains to build structural correspondences and generate an auxiliary task to predict the relation between any two adjacent words in the dependency tree.
- A unified framework of active transfer learning for cross-system recommendation85
A new transfer learning framework for recommender systems is proposed, which relaxes the above assumption to facilitate flexible knowledge transfer across different systems with low cost by using an active learning principle to construct entity correspondences across systems.
- 82
- 80
- Semantic-Discriminative Mixup for Generalizable Sensor-based Cross-domain Activity Recognition78
This paper introduces semantic-aware Mixup that considers the activity semantic ranges to overcome the semantic inconsistency brought by domain differences and introduces the large margin loss to enhance the discrimination of Mixup to prevent misclassification brought by noisy virtual labels.
- Distributed Multi-Task Relationship Learning75
This paper proposes a distributed multi-task learning framework that simultaneously learns predictive models for each task as well as task relationships between tasks alternatingly in the parameter server paradigm and proposes a communication-efficient primal-dual distributed optimization algorithm to solve theDual problem by carefully designing local subproblems to make the dual problem decomposable.
- 75
- Transferring localization models across space72
A novel approach to transferring the learning model trained on data from one area of a building to another by solving an optimization problem based on manifold learning techniques and can significantly reduce the amount of labeled data which are required for building the localization model.
- Riemannian Pursuit for Big Matrix Recovery67
An efficient method, called Riemannian Pursuit, that aims to address low rank matrix recovery and fixed-rank optimization problems simultaneously and substantially outperforms existing methods when applied to large-scale and ill-conditioned matrices.
- A Novel Distribution-Embedded Neural Network for Sensor-Based Activity Recognition64
A novel deep learning model is proposed to automatically learn meaningful features including statistical features, temporal features and spatial correlation features for activity recognition in a unified framework and is compared with state-of-the-art baselines.
- Multi-domain active learning for text classification64
This paper proposes a novel multi-domain active learning framework to jointly select data instances from all domains with duplicate information considered, and compares its method with the state-of-the-art active learning approaches on several text classification tasks.
- 62
- SC2Net: Sparse LSTMs for Sparse Coding61
This work recasts ISTA as a recurrent neural network unit and shows its connection with the well-known long short term memory (LSTM) model, and presents a neural network to achieve sparse codes in an end-to-end manner.
- Mining competent case bases for case-based reasoning60
This paper develops a theoretical framework for the error bound in case-based reasoning, and proposes a novel case-base mining algorithm guided by the theoretical results that returns a high-quality case base from raw data efficiently.
- 58
- Integrating Deep Learning with Logic Fusion for Information Extraction56
This work proposes to integrate logical knowledge in the form of first-order logic into a deep learning system, which can be trained jointly in an end-to-end manner and demonstrates the effectiveness and generalization of the proposed model on multiple IE tasks.
- Predicting user activity level in social networks56
A personalized and social regularized time-decay model for user activity level prediction is developed and experiments validate the effectiveness of the proposed model compared with some baselines including traditional supervised learning methods and node classification methods in social networks.
- Fast Graph Generation via Spectral Diffusion55
This paper argues that running full-rank diffusion SDEs on the whole graph adjacency matrix space hinders diffusion models from learning graph topology generation, and hence significantly deteriorates the quality of generated graph data, and proposes an efficient yet effective Graph Spectral Diffusion Model (GSDM), which is driven by low-rank diffusion SDEs on the graph spectrum space.
- Multi-class heterogeneous domain adaptation55
An efficient method named Sparse Heterogeneous Feature Representation (SHFR) is presented for multi-class HDA to learn a sparse feature transformation between domains with multiple classes and it is shown that the estimation error of the learned transformation decreases with the increasing number of binary classifiers.
- A deep learning framework for Hybrid Heterogeneous Transfer Learning55
This paper offers a new heterogeneous transfer learning framework named Hybrid Heterogeneous Transfer Learning (HHTL), which allows the selection of corresponding instances across domains to be biased to the source or target domain.
- Supervised Representation Learning with Double Encoding-Layer Autoencoder for Transfer Learning55
This article adapts the autoencoder technique to transfer learning and proposes a supervised representation learning method based on double encoding-layer autoen coder that can better reflect the degree of transfer difficulty and has stronger correlation with the performance from supervised learning algorithms.
- MetaQuant: Learning to Quantize by Learning to Penetrate Non-differentiable Quantization54
A meta network is trained using g_q and r as inputs, and outputs $g_r$ for subsequent weight updates, which alleviates the problem of non-differentiability, and can be trained in an end-to-end manner.
- 54
- Source Free Transfer Learning for Text Classification51
This paper proposes a transfer learning framework that effectively selects helpful auxiliary data from an open knowledge space and calls it Source Free Transfer Learning (SFTL), and highlights the automatic constructions of queries and the robustness of the SFTL framework.
- Transfer Learning for Cross-Language Text Categorization through Active Correspondences Construction50
This paper presents a general framework to integrate active learning to construct correspondences between heterogeneous domains for HTL, namely HTL through active correspondences construction (HTLA) and develops a new HTL method.
- Data Poisoning Attacks on Multi-Task Relationship Learning47
This paper focuses on multi-task relationship learning (MTRL) models, a popular subclass of MTL models where task relationships are quantized and are learned directly from training data and proposes an efficient algorithm called PATOM for computing optimal attack strategies.
- KVTuner: Sensitivity-Aware Layer-Wise Mixed-Precision KV Cache Quantization for Efficient and Nearly Lossless LLM Inference46
A simple yet effective framework KVTuner is proposed to adaptively search for the optimal hardware-friendly layer-wise KV quantization precision pairs for coarse-grained KV cache with multi-objective optimization and directly utilize the offline searched configurations during online inference.
- Transferable Interactive Memory Network for Domain Adaptation in Fine-Grained Opinion Extraction44
This work proposes an interactive memory network that consists of local and global memory units that could exploit both local andglobal memory interactions to capture intra-correlations among aspect words or opinion words themselves, as well as the interconnections between aspect and opinion words.
- 44
- Collaborative Sequential Recommendations via Multi-view GNN-transformers42
This work proposes a new sequential recommendation framework, which encodes the context information in each individual user behavior sequence as well as the collaborative information among the behavior sequences of different users, through building a local dependency graph for each item.
- Heterogeneous Transfer Learning via Deep Matrix Completion with Adversarial Kernel Embedding42
This paper proposes a new HTL method based on a deep matrix completion framework, where kernel embedding of distributions is trained in an adversarial manner for learning heterogeneous features across domains.
- Domain-constrained semi-supervised mining of tracking models in sensor networks41
A probabilistic semi supervised learning approach to reduce the calibration effort and increase the tracking accuracy and the results demonstrate the advantages of methods compared to other state-of-the-art object-tracking algorithms.
- Transfer joint embedding for cross-domain named entity recognition39
A new transfer learning method, named Transfer Joint Embedding (TJE), for cross-domain multiclass classification, which can fully exploit the relationships between classes (labels), and reduce domain difference in data distributions for transfer learning.
- 38
- Collaborative evolution for user profiling in recommender systems37
A novel evolutionary view of user's profile is proposed by proposing a Collaborative Evolution (CE) model, which learns the evolution ofuser's profiles through the sparse historical data in recommender systems and outputs the prospective user profile of the future.
- 36
- Deep Weighted MaxSAT for Aspect-based Opinion Extraction35
This model associates the benefits of high-level feature learning, knowledge reasoning, and structured learning with observable performance gain for the task of aspect-based opinion extraction and smoothly incorporate a weighted version of MaxSAT that connects deep neural networks and a graphical model in a joint framework.
- A Unified Event Coreference Resolution by Integrating Multiple Resolvers34
A globally optimized coreference resolution system dedicated to various sophisticated event coreference phenomena and an efficient and globally optimized graph partitioning model is employed for coreference chain formation using spectral partitioning which allows the incorporation of pronoun coreference information.
- 33
- Deep Neural Network Quantization via Layer-Wise Optimization Using Limited Training Data33
A novel layer-wise quantization method for deep neural networks, which only requires limited training data (1% of original dataset) and is proved that the final performance drop after quantization is bounded by a linear combination of the reconstructed errors caused at each layer.
- Syntactically Meaningful and Transferable Recursive Neural Networks for Aspect and Opinion Extraction33
This article constructs transferable recursive neural networks to automatically learn the domain-invariant fine-grained interactions among aspect words and opinion words and integrates the recursive neural network with a sequence labeling classifier on top that models contextual influence in the final predictions.
- Storage Efficient and Dynamic Flexible Runtime Channel Pruning via Deep Reinforcement Learning32
Comparison experimental results with existing runtime and static pruning methods on state-of-the-art CNNs demonstrate that the proposed framework is able to provide a tradeoff between dynamic flexibility and storage efficiency in runtime channel pruning.
- Learning an Adaptive Meta Model-Generator for Incrementally Updating Recommender Systems29
This work proposes a novel Adaptive Sequential Model Generation (ASMG) framework, which generates a better serving model from a sequence of historical models via a meta generator, and proposes to employ Gated Recurrent Units (GRUs) to leverage its ability to capture the long-term dependencies.
- 28
- Transfer Learning with Multiple Sources via Consensus Regularized Autoencoders28
In the proposed framework, autoencoders are adopted to construct a feature mapping from an original instance to a hidden representation, and train multiple classifiers from the source domain data jointly by performing an entropy-based consensus regularizer on the predictions on the target domain.
- 28
- Multi-view Positive and Unlabeled Learning27
This work proposes a novel PU learning method based on density ratio estimation without constructing any sets of negative instances or estimating any intermediate densities, and extends the proposed learning method in a multi-view manner by utilizing multiple heterogeneous sources.
- Online co-localization in indoor wireless networks by dimension reduction27
An online co-localization method that can deal with labeled and unlabeled data stream based on semi-supervised manifold-learning techniques is proposed that can achieve high accuracy with less calibration effort as compared to several previous systems.
- Backdoor Attacks on Dense Retrieval via Public and Unintentional Triggers26
A novel scenario where the attackers aim to covertly disseminate targeted misinformation through a retrieval system through a retrieval system triggered by grammar errors in dense passage retrieval is introduced.
- Communication-Efficient Distributed PCA by Riemannian Optimization24
A communication-efficient algorithm based on Riemannian optimization, which trades local computation for global communication, which achieves a significant reduction of communication cost over existing distributed PCA algorithms is proposed.
- Transition-based Adversarial Network for Cross-lingual Aspect Extraction24
A novel deep model to transfer knowledge from a source language with labeled training data to a target language without any annotations is developed and achieves state-of-the-art performance on English, French and Spanish restaurant review datasets.
- 23
- Weakly-supervised sensor-based activity segmentation and recognition via learning from distributions23
A unified weakly-supervised framework based on kernel embedding of distributions to jointly segment sensor streams, extract powerful features from each segment, and train a final classifier for activity recognition is proposed.
- 21
- Deep Inductive Logic Reasoning for Multi-Hop Reading Comprehension20
A deep-learning based inductive logic reasoning method that firstly extracts query-related (candidate-related) information, and then conducts logic reasoning among the filtered information by inducing feasible rules that entail the target relation is proposed.
- Memory networks for fine-grained opinion mining20
This work develops an end-to-end multi-task memory network, where aspect/opinion terms extraction for a specific category is considered as a task, and all the tasks are learned jointly by exploring commonalities and relationships among them.
- 20
- Learning Representations Via a Robust Behavioral Metric for Deep Reinforcement Learning19
A new behavioral distance is proposed, the RAP distance, and a practical representation learning algorithm is developed on top of it with a theoretical analysis based on the potential relaxation and/or approximation gaps for existing behavioral metric-based representation learning methods.
- Cooperative Pruning in Cross-Domain Deep Neural Network Compression19
This paper proposes a method to perform crossdomain pruning by cooperatively training in both domains: taking advantage of data and a pre-trained model from the source domain to assist pruning in the target domain.
- 17
- Heterogeneous Multi-task Semantic Feature Learning for Classification17
This paper proposes a multi-task nonnegative matrix factorization (MTNMF) method to learn a common semantic feature space underlying different heterogeneous feature spaces of each task and model the heterogenous MTL problem as a multiple-task multi-view learning (MTMVL) problem.
- Extended Spectral Regression for efficient scene recognition17
A new SR approach, called Extended Spectral Regression (ESR), is proposed to perform manifold learning on a huge number of data samples and an efficient Bag-of-Words (BOW) based method is developed which employs ESR to encapsulate local visual features with their semantic, spatial, scale, and orientation information for scene recognition.
- Learning recursive HTN-method structures for planning17
This work presents a probabilistic model for unsupervised learning of HTN methods from action sequences and introduces a novel two-pronged approach by simultaneously learning a Markov model for action segment clusters fromaction sequences and then learning an action parameter model for recognizing tasks.
- Trustworthy Machine Learning: Robustness, Generalization, and Interpretability16
This tutorial will give a comprehensive introduction to the recent advance of trustworthy machine learning in robustness, generalization, and interpretability.
- 16
- 16
- Does In-Context Learning Really Learn? Rethinking How Large Language Models Respond and Solve Tasks via In-Context Learning15
It is demonstrated that retrieving the semantically similar examples notably boosts the model's discriminative capability, and a trade-off in selecting good in-context examples regarding label diversity is observed in selecting good in-context examples regarding label diversity.
- 15
- Adaptive Group Sparse Multi-task Learning via Trace Lasso15
A new MTL method that can adaptively group correlated tasks into clusters and share information among the correlated tasks only is proposed, based on the assumption that each task parameter is a linear combination of other tasks’ and the coefficients of the linear combination are active only if there is relatedness between the two tasks.
- Adversarial Transfer Learning14
A survey of the field of GANs, looking at the original formulation, training variants, evaluation methods, and extensions, and recent work on transfer learning, focusing on comparing different adversarial domain adaptation methods.
- Counting via LED sensing: Inferring occupancy using lighting infrastructure14
CeilingSee is a dedicated occupancy inference system free of heavy infrastructure deployments and user involvements that converts part of the ceiling-mounted LED luminaires to act as sensors, sensing the variances in diffuse reflection caused by occupants.
- 13
- Distribution-Based Semi-Supervised Learning for Activity Recognition13
This paper proposes a novel method, named Distribution-based Semi-Supervised Learning, capable of automatically extracting powerful features with no domain knowledge required, meanwhile, alleviating the heavy annotation effort through semi-supervised learning.
- Class-wise supervised hashing with label embedding and active bits13
A class-wise supervised hashing method that trains a model based on a class-pairwise similarity matrix, whose size is much smaller than the instance-pair wise similarity matrix in many applications.
- Preprocessing Time Series Data for Classification with Application to CRM12
The theory and algorithms for the task are provided, and it is empirically demonstrated that the method is effective in determining the customer class for CRM applications in the telecommunications industry.
- 11
- 11
- 11
- 10
- Variational Deep Logic Network for Joint Inference of Entities and Relations9
A variational deep logic network that incorporates both representation learning and relational reasoning via the variational EM algorithm is proposed that consists of a deep neural network to learn high-level features with implicit interactions via the self-attention mechanism and a relational logic network to explicitly exploit target interactions.
- Sequential Generative Exploration Model for Partially Observable Reinforcement Learning9
A novel reward shaping approach to infer the intrinsic rewards for the agent from a sequential generative model that is able to derive intrinsic rewards that could better reflect the agent's surprise or curiosity over its ground-truth state by taking a sequential inference procedure.
- The Pareto Principle Is Everywhere: Finding Informative Sentences for Opinion Summarization Through Leader Detection9
The informative sentence selection problem in opinion summarization is formulated as a community leader detection problem, where a community consists of a cluster of sentences toward the same aspect of an entity and leaders can be considered as the most informative sentences of the corresponding aspect.
- Overcoming Negative Transfer by Online Selection: Distant Domain Adaptation for Fault Diagnosis8
This work coins the term “distant domain adaptation problem” to describe the challenge of adapting from a labeled source domain to a significantly disparate unlabeled target domain, and develops a class-conditional adversarial adaptation to address the label distribution disparities while learning domain invariant representation to account for potential label distribution disparities between the domains.
- Learning Generalizable Representations for Reinforcement Learning via Adaptive Meta-learner of Behavioral Similarities8
A novel meta-learner-based framework for representation learning regarding behavioral similarities for reinforcement learning that outperforms state-of-the-art baselines on several benchmarks, including conventional DM Control Suite, Distracting DM Control suite and a self-driving task CARLA.
- Multi-task Coupled Attentions for Category-specific Aspect and Opinion Terms Co-extraction.8
This work proposes an end-to-end multi-task attention model, where each task corresponds to aspect/opinion terms extraction for a specific category, and demonstrates its state-of-the-art performance on three benchmark datasets.
- 8
- 8
- Transfer learning with applications on text, sensors and images8
This talk will first give an overview of transfer learning, and then present three specific transfer learning methods with applications to opinion mining from text, image classification using auxiliary text information, and indoor Wifi-based localization.
- Graph Principal Flow Network for Conditional Graph Generation7
This work proposes a novel graph conditional generative model, Graph Principal Flow Network (GPrinFlowNet), which enables to progressively generate high-quality graphs from low- to high-frequency components for a given graph label.
- 7
- Transfer Value Iteration Networks7
The proposed TVINs outperform VINs when the source and the target domains have similar but not identical action and feature spaces, and is consistent across different environments, maze sizes, dataset sizes as well as different values of hyperparameters such as number of iteration and kernel size.
- 7
- Analytical FFN-to-MoE Restructuring via Activation Pattern Analysis6
An analytical post-training framework that rapidly restructures FFNs into sparse MoE architectures using only a small calibration dataset is proposed, outperforming methods requiring orders of magnitude more resources.
- A Diffusion Model with State Estimation for Degradation-Blind Inverse Imaging6
This work proposes a learnable state-estimator-based diffusion model to incorporate the measurements into the reconstruction process and makes efficient use of the pre-trained diffusion models with computational feasibility compared to the conditional diffusion models, which need to be trained from scratch.
- 6
- Ensemble of Anchor Adapters for Transfer Learning6
An ENsemble framework of anCHOR adapters (ENCHOR for short), in which an anchor adapter adapts the features of instances based on their similarities to a specific anchor (i.e., a selected instance), which greatly alleviates negative transfer.
- 6
- Learning gradient-based mixup with extrapolation toward flatter minima for domain generalization5
This work carefully design a policy to generate the instance weights, named Flatness-aware Gradient-based Mixup (FGMix), which relies on gradient-based compatibilities to assign greater weights to instances that carry more invariant information and learn the mixup policy towards flatter minima for better generalization.
- 5
- 5
- Hashing over Predicted Future Frames for Informed Exploration of Deep Reinforcement Learning5
This paper builds the capability for an RL agent to predict over the future transitions and evaluate the frequentness for the predicted future frames in a meaningful manner and derives a reliable metric for evaluating the novelty of the future direction pointed by each action, and informs the agent to explore the least frequent one.
- Mitigating Performance Saturation in Neural Marked Point Processes: Architectures and Loss Functions4
Experimental results show that GCHP can significantly reduce training time and the likelihood ratio loss with interarrival time probability assumptions can greatly improve the model performance, and a simple graph-based network structure called G CHP is proposed, which utilizes only graph convolutional layers, thus it can be easily accelerated by the parallel mechanism.
- Towards Faster Deep Collaborative Filtering via Hierarchical Decision Networks4
This paper proposes a new deep learning-based hierarchical decision network to filter out irrelevant items to save computation cost while maintaining good recommendation accuracy of deep CF methods and develops a distillation-based training algorithm, which uses a well-trained CF model as a teacher network to guide the training of the decision network.
- Communication-efficient distributed multi-task learning with matrix sparsity regularization4
This work proposes a fast communication-efficient distributed optimization method for solving the problem of multi-task learning with matrix sparsity regularization, and theoretically proves that it enjoys a fast convergence rate for different types of loss functions in the distributed environment.
- Beyond Speedup -- Utilizing KV Cache for Sampling and Reasoning3
This work proposes treating the KV cache as a lightweight representation, eliminating the need to recompute or store full hidden states, opening new directions for representation reuse in LLM inference.
- 3
- 3
- Domain Adaptive Semantic Segmentation Through Structure Enhancement3
This work proposes an effective method to adapt the segmentation network trained on synthetic images to real scenarios in an unsupervised fashion and enhances the structure information of the target images at both the feature level and the output level to improve the adaptation performance for semantic segmentation.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 1
- Improving the Generalization of Unseen Crowd Behaviors for Reinforcement Learning based Local Motion Planners1
This work introduces an efficient method that enhances agent diversity within a single policy by maximizing an information-theoretic objective and proposes diverse scenarios inspired by pedestrian crowd behaviors that improve an agent’s robustness against unseen crowds.
- 1
- A Tutorial on Domain Generalization1
This tutorial presents the recent advance of domain generalization and introduces the background, formulation, and theory behind this topic, with a primary focus on the methodology, evaluation, and applications.
- 1
- 1
- Transfer Learning in Recommender Systems1
Experimental results show that CDs & Vinyl recommendations can indeed benefit from making use of Digital Music recommendations.
- 1
- 1
- 1
- A Sequential Condition Evolution and Interaction Knowledge Graph Framework for Traditional Chinese Medicine Recommendation–
This paper proposes a novel approach for optimizing TCM recommendations based on Sequential Condition Evolution and Interaction Knowledge Graphs SCEIKG, an innovation framework that conceptualizes the method as a sequential prescription-making problem by considering the inherent dynamism of patients’ conditions across multiple diagnoses.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Communication-Efficient Distributed Primal-Dual Algorithm for Saddle Point Problem.–
This paper proposes a novel communication-efficient distributed optimization framework to solve the convex-concave saddle point problem based on primal-dual methods, and provides a convergence analysis of the proposed algorithm, and extends it to address non-smooth and non-strongly convex loss functions.
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.