Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Suggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works50 from public data
- Named entity recognition337
It is shown that the maximum entropy framework is able to make use of global information directly, and achieves performance that is comparable to the best previous machine learning-based NERs on M UC-6 and MUC-7 test data.
- Named entity recognition with a maximum entropy approach275
The named entity recognition (NER) task involves identifying noun phrases that are names, and assigning a class to each name.
- Interpretable Rumor Detection in Microblogs by Attending to User Interactions245
This work proposes a post-level attention model (PLAN) to model long distance interactions between tweets with the multi-head attention mechanism in a transformer network and is the first to evaluate the models on two rumor detection data sets: the PHEME data set as well as the Twitter15 and Twitter16 data sets.
- Query based event extraction along a timeline176
A framework and a system that extracts events relevant to a query from a collection C of documents, and places such events along a timeline, based on the assumption that "important" events are widely cited in many documents for a period of time within which these events are of interest.
- Bayesian online classifiers for text classification and filtering160
Empirical results indicate that these Bayesian online classifiers are comparable with the best text classification systems and the online approach offers the advantage of continuous learning in the batch-adaptive text filtering task.
- 152
- Optimizing F-measures: A tale of two approaches136
The theoretical justifications and connections for these two approaches for learning to maximize F-measures are investigated, and the conditions under which one approach is preferable to the other using synthetic and real datasets are studied.
- Language modeling with sum-product networks83
It is shown that the ability of SPN to use hidden layers to model complex dependencies among words, and its tractable inference and learning times, make it a suitable framework for a language model.
- 79
- Closing the gap71
This is the first research to have demonstrated that a learning approach to the full-scale information extraction task could achieve performance rivaling that of the knowledge engineering approach.
- Domain adaptive bootstrapping for named entity recognition62
This paper proposes an effective domain adaptive bootstrapping algorithm that selects unlabeled target domain data that are informative about the target domain and easy to automatically label correctly, and shows that the method outperforms supervised, transductive and bootstrapped algorithms on the named entity recognition task.
- Conditional random field with high-order dependencies for sequence labeling and segmentation58
Effcient inference algorithms are given to handle high-order dependencies between labels or segments in conditional random fields, under the assumption that the number of distinct label patterns used in the features is small, which leads to effcient learning algorithms for these conditional Random fields.
- Conditional Random Fields with High-Order Features for Sequence Labeling52
It is shown that it is possible to design efficient inference algorithms for a conditional random field using features that depend on long consecutive label sequences (high-order features), as long as the number of distinct label sequences used in the features is small.
- Coupled Hierarchical Transformer for Stance-Aware Rumor Verification in Social Media Conversations50
To extend BERT to obtain thread representations, this paper proposes a Hierarchical Transformer, which divides each long thread into shorter subthreads, and employs BERT to separately represent each subthread, followed by a global Transformer layer to encode all the subthreads.
- Can Syntax Help? Improving an LSTM-based Sentence Compression Model for New Domains45
This paper proposes two major changes to the LSTM neural network model for sentence compression: using explicit syntactic features and introducing syntactic constraints through Integer Linear Programming (ILP).
- Measuring and Enhancing Trustworthiness of LLMs in RAG through Grounded Attributions and Learning to Refuse40
The results show that various prompting methods, such as in-context learning, fail to effectively adapt LLMs to the RAG task as measured by Trust-Score, and proposes Trust-Align, a method to align LLMs for improved Trust-Score performance.
- 39
- Active Learning for Probabilistic Hypotheses Using the Maximum Gibbs Error Criterion37
The experimental results on a named entity recognition task and a text classification task show that the maximum Gibbs error criterion is an effective active learning criterion for noisy models.
- Mitigating Misinformation in Online Social Network with Top-k Debunkers and Evolving User Opinions36
A new opinion model is proposed that takes into account users’ biases and their social neighbours’ opinions and design a mitigation solution to identify a subset of debunkers that maximizes the number of users who have been exposed to the misinformation but choose to believe the counter message.
- Activity Recognition from Physiological Data using Conditional Random Fields34
It is shown that linear chain CRF can effectively make use of the sequential information in the data, and, with Expectation Maximization, can be trained on partially unlabeled sessions to improve performance.
- Universal Dependencies Parsing for Colloquial Singaporean English32
This work investigates dependency parsing of Singlish by constructing a dependency treebank under the Universal Dependencies scheme, and then training a neural network model by integrating English syntactic knowledge into a state-of-the-art parser trained on the Singlish treebank.
- Extracting Relation Descriptors with Conditional Random Fields32
This paper proposes two modifications to linear-chain CRFs, namely, reducing the space of possible label sequences and introducing long-range features, and finds that both modifications can significantly improve the performance for this novel relation extraction problem.
- Learning to Capitalize with Character-Level Recurrent Neural Networks: An Empirical Study27
An approach using character-level recurrent neural networks (RNN), which performs competitively compared to language modeling and conditional random fields (CRF) approaches is proposed.
- Twitter Homophily: Network Based Prediction of User’s Occupation24
This study extends an existing data set for this problem, and achieves significantly better performance by using social network homophily that has not been fully exploited in previous work by using the graph convolutional network to exploit socialhomophily.
- Robust Domain Adaptation for Relation Extraction via Clustering Consistency19
A two-phase framework to adapt existing relation extraction classifiers to extract relations for new target domains to address two challenges: negative transfer when knowledge in source domains is used without considering the differences in relation distributions; and lack of adequate labeled samples for rarer relations in the new domain, due to a small labeled data set and imbalance relation distributions.
- Domain Adaptation for Coreference Resolution: An Adaptive Ensemble Approach18
This work is the first to both develop a domain adaptation algorithm for the coreference resolution problem and have the above features as an ensemble method.
- A General Regularization Framework for Domain Adaptation17
A domain adaptation framework is proposed, and it is formally proved that it generalizes the feature augmentation technique in and the multi-task regularization framework in and allows practitioners to tune hyper-parameters to encourage transfer between close domains and avoid negative transfer between distant ones.
- Teaching a weaker classifier17
This paper describes how a machine-learning named entity recognizer (NER) on upper case text can be improved by using a mixed case NER and some unlabeled text, which reduces the performance gap between the mixed caseNER and the upper case N ER substantially.
- Co-training for Commit Classification16
This paper applies co-training, a semi-supervised learning method, to take advantage of the two views available – the commit message and the code changes – to improve commit classification.
- Cross-Topic Rumor Detection using Topic-Mixtures14
This paper proposes a method inspired by the “mixture of experts” paradigm that can outperform two baseline debiasing methods in a cross-topic setting and works better than the baselines, showing that the method does not rely on superficial features.
- Unsupervised Information Extraction with Distributional Prior Knowledge9
A generative model that incorporates distributional prior knowledge to help distribute candidates in a document into appropriate slots is proposed that can bring substantial improvements to the task of automatic discovery of information extraction template from a given text collection.
- Lessons from Training Grounded LLMs with Verifiable Rewards8
This work uses the GRPO (Group Relative Policy Optimization) method to train models using verifiable outcome-based rewards targeting answer correctness, citation sufficiency, and refusal quality, without requiring gold reasoning traces or expensive annotations.
- 8
- Cooled and Relaxed Survey Propagation for MRFs8
A new algorithm, Relaxed Survey Propagation (RSP), for finding MAP configurations in Markov random fields shows that it outperforms all approaches for Ising models with mixed couplings, as well as on a web person disambiguation task formulated as a supervised clustering problem.
- 7
- 5
- 3
- Intrinsic evaluation of language models for code-switching2
This paper first put this assumption into question, and observe that alternatively generated sentences could often be linguistically correct when they differ from the ground truth by only one edit, and showed that by using multi-lingual BERT, this paper can achieve better performance than previous work on two code-switching data sets.
- 2
- Relaxed survey propagation: a sum-product algorithm for Max-SAT2
RSP is adapted to solve Max-SAT problems, and it is shown that it has an interpretation of estimating marginals over covers violating a minimum number of clauses, which naturally generalizes the cover interpretation of SP.
- Tokenizer-Agnostic Engram Module1
This work replaces the XOR-based hashing with the general polynomial hashing with a joint embedding space across $N$, and achieves tokenizer-agnosticism: hash equivalence for byte-equivalent token sequences.
- Colloquial Singaporean English Style Transfer with Fine-Grained Explainable Control1
This work constructs a large, high-quality dataset of formal and informal sentences, annotated across six linguistic aspects, and proposes a novel multi-agent framework where large language models (LLMs) act as expert agents for each linguistic aspect.
- Finite Energy Survey Propagation for Constraint Satisfaction Problems1
The SP-Y algorithm for general constraint satisfaction problems, applicable for minimizing the number of violated constraints is formulated, and preliminary results show that SP can solve a few instances of induced subgraph isomorphism for which belief propagation failed to converge.
- Spokes: Optimizing for Diverse Pretraining Data Selection–
This work directly optimize diversity by introducing a probabilistic diversification framework based on the G-Vendi score, optimized via exponentiated gradient descent, which produces subsets that are substantially more diverse than those obtained via random sampling.
- –
- –
- –
- –
- –
- Supplementary Material to "Cooled and Relaxed Survey Propagation for MRFs"–
This is the supplementary material for the submission to NIPS 2007, entitled “Cooled and Relaxed Survey Propagation for MRFs”, to prove the update equations of relaxationpropagation (RSP) in the main paper.
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.