Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks77 from public data
- ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate1,015
A multi-agent referee team called ChatEval is constructed to autonomously discuss and evaluate the quality of generated responses from different models on open-ended questions and traditional natural language generation (NLG) tasks, offering a human-mimicking evaluation process for reliable assessments.
- MentalBERT: Publicly Available Pretrained Language Models for Mental Healthcare485
This paper trains and releases two pretrained masked language models, i.e., MentalBERT and MentalRoBERTa, to benefit machine learning for the mental healthcare research community and demonstrates that language representations pretrained in the target domain improve the performance of mental health detection tasks.
- Pre-trained Language Models in Biomedical Domain: A Systematic Survey256
The motivations of PLMs in the biomedical domain are discussed and the key concepts of pre-trained language models are introduced and a taxonomy of existing biomedical PLMs is proposed that categorizes them from various perspectives systematically.
- RQ-RAG: Learning to Refine Queries for Retrieval Augmented Generation213
This paper proposes learning to Refine Query for Retrieval Augmented Generation (RQ-RAG), endeavoring to enhance the model by equipping it with capabilities for explicit rewriting, decomposition, and disambiguation, and demonstrates enhanced performance in handling complex, multi-hop QA datasets.
- AffectiveSpace 2: Enabling Affective Intuition for Concept-Level Sentiment Analysis192
In conclusion, AffectiveSpace 2 is a vector space model, built by means of random projection, that allows for reasoning by analogy on natural language speech and enables the inference of emotions and polarity conveyed by multi-word expressions, thus achieving efficient concept-level sentiment analysis.
- Graph Neural Networks with Generated Parameters for Relation Extraction176
A novel graph neural network with generated parameters (GP-GNNs), i.e. the transition matrices used in message passing procedure, are produced by a generator taking natural language sentences as inputs, to demonstrate that multi-hop reasoning mechanism yields significant improvements.
- 162
- Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models158
This study proposes ProAttack, a novel and efficient method for performing clean-label backdoor attacks based on the prompt, which uses the prompt itself as a trigger, which does not require external triggers and ensures correct labeling of poisoned samples, improving the stealthy nature of the backdoor attack.
- A Survey of Reasoning with Foundation Models: Concepts, Methodologies, and Outlook142
This article introduces seminal foundation models proposed or adaptable for reasoning, highlighting the latest advancements in various reasoning tasks, methods, and benchmarks, and delves into the potential future directions behind the emergence of reasoning abilities within foundation models.
- Simple and Effective Curriculum Pointer-Generator Networks for Reading Comprehension over Long Narratives126
A curriculum learning (CL) based Pointer-Generator framework for reading/sampling over large documents, enabling diverse training of the neural model based on the notion of alternating contextual difficulty, and a new Introspective Alignment Layer (IAL), which reasons over decomposed alignments using block-based self-attention.
- GIMLET: A Unified Graph-Text Model for Instruction-Based Molecule Zero-Shot Learning97
Experimental results demonstrate that GIMLET significantly outperforms molecule-text baselines in instruction-based zero-shot learning, even achieving closed results to supervised GNN models on tasks such as toxcast and muv.
- GLGE: A New General Language Generation Evaluation Benchmark75
The General Language Generation Evaluation (GLGE), a new multi-task benchmark for evaluating the generalization capabilities of NLG models across eight language generation tasks, is presented and a leaderboard with strong baselines including MASS, BART, and ProphetNet is built.
- Lightweight and Efficient Neural Natural Language Processing with Quaternion Networks72
This paper proposes a series of lightweight and memory efficient neural architectures for a potpourri of natural language processing (NLP) tasks, exploiting computation using Quaternion algebra and hypercomplex spaces, enabling not only expressive inter-component interactions but also significantly reduced parameter size due to lesser degrees of freedom in the Hamilton product.
- 61
- RikiNet: Reading Wikipedia Pages for Natural Question Answering59
This paper introduces a new model, called RikiNet, which reads Wikipedia pages for natural question answering, which is the first single model that outperforms the single human performance.
- AI Alignment: A Contemporary Survey54
This survey aims to provide a comprehensive yet beginner-friendly review of alignment research topics and discusses current and prospective governance practices adopted by governments, industry actors, and other third parties, aimed at managing existing and future AI risks.
- Revision in Continuous Space: Unsupervised Text Style Transfer without Adversarial Learning53
A new framework that utilizes the gradients to revise the sentence in a continuous space during inference to achieve text style transfer and significantly outperforms five state-of-the-art methods.
- 50
- Interactive Language Learning by Question Answering48
This work proposes and evaluates a set of baseline models for the QAit task that includes deep reinforcement learning agents, and shows that the task presents a major challenge for machine reading systems, while humans solve it with relative ease.
- Tracking single-cell evolution using clock-like chromatin accessibility loci41
A method that counts the fraction of opened clock-like loci from scATAC-seq data to determine cell age and perform lineage tracing in various cell lineages and animal species and shows concordance with known developmental hierarchies, correlates well with DNAm-based clocks and is complementary with mutation-based lineage tracing, RNA velocity and stemness predictions.
- Tell Me How to Ask Again: Question Data Augmentation with Controllable Rewriting in Continuous Space40
A novel data augmentation method, referred to as Controllable Rewriting based Question Data Augmentation (CRQDA), for machine reading comprehension, question generation, and question-answering natural language inference tasks.
- 38
- 36
- Few-Shot Domain Adaptation with Polymorphic Transformers28
A Polymorphic Transformer (Polyformer), which can be incorporated into any DNN backbones for few-shot domain adaptation, and which can perform robustly on the target domain after being trained on a few annotated images.
- On Orthogonality Constraints for Transformers24
This paper studies orthogonality constraints for transformers, showing the effectiveness with empirical evidence from ten machine translation tasks and two dialogue generation tasks and close to the 29.7 BLEU achieved by the very competitive dynamic convolution.
- Would you Rather? A New Benchmark for Learning Machine Alignment with Cultural Values and Social Preferences22
A newly introduced problem concerned with predicting the preferable options from two sentences describing scenarios that may involve social and cultural situations, framed as a natural language inference task with crowd-sourced preference votes by human players, obtained from a gamified voting platform.
- Diverse, Controllable, and Keyphrase-Aware: A Corpus and Method for News Multi-Headline Generation22
A multi-source Transformer decoder, which takes three sources as inputs and generates multiple headlines with keyphrases of user interests to generate multiple keyphrase-relevant headlines, achieves state-of-the-art results in terms of quality and diversity.
- 21
- 20
- Interactive Machine Comprehension with Information Seeking Agents20
This paper “occlude” the majority of a document’s text and add context-sensitive commands that reveal “glimpses” of the hidden text to a model, and believes that this setting can contribute in scaling models to web-level QA scenarios.
- Learning Multi-Task Communication with Message Passing for Sequence Learning20
This work adopts the idea from message-passing graph neural networks, and proposes a general graph multi-task learning framework in which different tasks can communicate with each other in an effective and interpretable way.
- ReForm-Eval: Evaluating Large Vision Language Models via Unified Re-Formulation of Task-Oriented Benchmarks17
This work proposes to re-formulate existing benchmarks into unified LVLM-compatible formats and presents ReForm-Eval benchmark, offering substantial data for evaluating various capabilities of LVLMs, demonstrating the comprehensiveness and reliability of ReForm-Eval in assessing various LVLMs.
- 17
- 16
- 15
- 15
- 13
- Revision in Continuous Space: Fine-Grained Control of Text Style Transfer.12
A new framework without adversarial learning is proposed, which consists of a variational auto-encoder (VAE), some attribute predictors (one for each attribute), and a content predictor and significantly outperforms five state-of-the-art methods.
- Concurrent multi‐agent systems with temporal logic objectives: game theoretic analysis and planning through negotiation12
A negotiation protocol is developed which ensures that under a proper design of preferences and tasks, the mutually accepted plan is a Pareto optimal pure Nash equilibrium.
- 10
- Think Before You Act: Decision Transformers with Working Memory10
This work proposes an internal working memory module to store, blend, and retrieve information for different downstream tasks in large language model-based decision-making agents, and demonstrates that memory fine-tuning further enhances the adaptability of the proposed architecture.
- 10
- VeriEquivBench: An Equivalence Score for Ground-Truth-Free Evaluation of Formally Verifiable Code9
The results show that generating formally verifiable code remains a profound challenge for state-of-the-art LLMs, and underscores both the difficulty of the task and the need for benchmarks like VeriEquivBench to drive progress toward scalable and reliable coding agents.
- Symbolic planning and control using game theory and grammatical inference9
A system can accomplish an objective specified in temporal logic while interacting with an unknown, dynamic but rule-governed environment, by employing grammatical inference and adapting its plan of action on-line to satisfy the task specification.
- 8
- Distilling Reverse-Mode Automatic Differentiation (DrMAD) for Optimizing Hyperparameters of Deep Neural Networks8
This work proposes to distill the knowledge of the forward pass into an shortcut path, through which the training trajectory is approximately reverse, making it feasible, for the first time, to automatically tune hundreds of thousands of hyperparameters of deep neural networks in practice.
- Integrating Grammatical Inference into Robotic Planning8
A grammatical inference module is incorporated that identifies the governing dynamics of the adversarial environment and game theory is utilized to compute a motion plan for a system given a task specification.
- Omnireg-gpt: a high-efficiency foundation model for comprehensive genomic sequence understanding7
OmniReg-GPT, a generative foundation model designed for the low-resource pretraining of long genomic sequences by optimized attention mechanism, is introduced with a hybrid local-global attention architecture, enabling efficient analysis of multi-scale regulatory features across long DNA sequences.
- 7
- PCEvolve: Private Contrastive Evolution for Synthetic Dataset Generation via Few-Shot Private Data and Generative APIs6
A novel API-assisted algorithm, Private Contrastive Evolution (PCEvolve), which iteratively mines inherent inter-class contrastive relationships in few-shot private data beyond individual data points and seamlessly integrates them into an adapted Exponential Mechanism to optimize DP's utility in an evolution loop is proposed.
- 6
- 5
- 5
- 5
- 4
- 3
- Information-Theoretic Opacity-Enforcement in Markov Decision Processes3
The paper develops primal-dual policy gradient methods for opacity-enforcement planning subject to constraints on total returns and proposes novel algorithms to compute the policy gradient of entropy for each observation, leveraging message passing within the hidden Markov models.
- 2
- Planning with Probabilistic Opacity and Transparency: A Computational Model of Opaque/Transparent Observations2
This paper develops an approach that uses the observation function, the secret, and the model of the dynamical systems to construct a so-called opaqueobservations automaton that accepts the exact set of observations that enforce opacity.
- Input Conditioned Graph Generation for Language Agents2
This research aims to develop both learnable and dynamic agents by using an existing framework that abstracts language agents as graphs to generate edges that represent the flow of communication within the graph based on the given input, thereby adjusting the internal communication of a language agent.
- 2
- 2
- 2
- DP-DCAN: Differentially Private Deep Contrastive Autoencoder Network for Single-cell Clustering2
A Differentially Private Deep Contrastive Autoencoder Network (DP-DCAN) by partial network perturbation for single-cell clustering is designed, which demonstrates strong robustness to adversarial attacks.
- 2
- 2
- 2
- Deep Reinforcement Learning for Accelerating the Convergence Rate2
This work proposes a principled deep reinforcement learning approach that is able to accelerate the convergence rate of general deep neural networks (DNNs) and uses deep RL to learn how to optimize a large-sized DNN.
- HUB: Guiding Learned Optimizers with Continuous Prompt Tuning1
This work proposes a hybrid-update-based (HUB) optimization strategy inspired by recent advancements in hard prompt tuning and result selection techniques used in large language and vision models that is able to retain the benefits of learned optimizers while stabilizing the training process and improving testing performance.
- 1
- Tracking single cell evolution via clock-like chromatin accessibility1
A simple, accurate and phenotypic-neutral measure of cell developmental hierarchy – the fraction of accessible clock-like loci from scATAC data to determine cell age and perform lineage tracing is shown.
- –
- –
- –
- –
- Learning semantic topics for domain-adapted textual knowledge transfer–
This work calculates the topic similarity between source and target domain, which is severed as an effective constraint to regularize the training process and hence improve the generalization of the source model to the target domain.
- The Design and Implementation of the Wireless Data Modem Base on TTIB Technology–
The theory of TTIB and FFSR is described, and a design of the data modem is provided by using the digital signal processing technology to implement TTIB-FFSR.
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.