Kai Wei

Ke Wang, Yuning Wu, Haoran Liu et al.

Self-distilled policy optimization (SDPO) has become a popular paradigm for LLM post-training, where a model learns from its own predictions conditioned on privileged information. SDPO, however, is sensitive to how much each update step should be trusted: corrections from a self-teacher can be highly informative on some batches and misleading on others, and applying them uniformly with a fixed step size can destabilize training. Drawing inspiration from viscous-fluid dynamics and formalizing the analogy at the SDE level, we propose Physics-Guided Policy Optimization (PGPO), which introduces an information-modulated step-size multiplier derived from a mutual-information estimate between the student's predictions and the feedback-conditioned teacher. We show that this modulation preserves the order-1 weak-approximation guarantees of vanilla SGD, and incurs negligible overhead per iteration. We evaluate PGPO on the Science-QA dataset, where it outperforms SDPO on 3 of the 4 domains with gains of up to +4.5 points, while remaining stable in a setting where SDPO collapses late in training.

Xiaolong Qian, Qi Jiang, Yao Gao et al.

Controllable Depth-of-Field (DoF) imaging commonly produces amazing visual effects based on heavy and expensive high-end lenses. However, confronted with the increasing demand for mobile scenarios, it is desirable to achieve a lightweight solution with Minimalist Optical Systems (MOS). This work centers around two major limitations of MOS, i.e., the severe optical aberrations and uncontrollable DoF, for achieving single-lens controllable DoF imaging via computational methods. A Depth-aware Controllable DoF Imaging (DCDI) framework is proposed equipped with All-in-Focus (AiF) aberration correction and monocular depth estimation, where the recovered image and corresponding depth map are utilized to produce imaging results under diverse DoFs of any high-end lens via patch-wise convolution. To address the depth-varying optical degradation, we introduce a Depth-aware Degradation-adaptive Training (DA2T) scheme. At the dataset level, a Depth-aware Aberration MOS (DAMOS) dataset is established based on the simulation of Point Spread Functions (PSFs) under different object distances. Additionally, we design two plug-and-play depth-aware mechanisms to embed depth information into the aberration image recovery for better tackling depth-aware degradation. Furthermore, we propose a storage-efficient Omni-Lens-Field model to represent the 4D PSF library of various lenses. With the predicted depth map, recovered image, and depth-aware PSF map inferred by Omni-Lens-Field, single-lens controllable DoF imaging is achieved. Comprehensive experimental results demonstrate that the proposed framework enhances the recovery performance, and attains impressive single-lens controllable DoF imaging results, providing a seminal baseline for this field. The source code and the established dataset will be publicly available at https://github.com/XiaolongQian/DCDI.

Kai Wei, Dillon Knox, Martin Radfar et al.

Dialogue act classification (DAC) is a critical task for spoken language understanding in dialogue systems. Prosodic features such as energy and pitch have been shown to be useful for DAC. Despite their importance, little research has explored neural approaches to integrate prosodic features into end-to-end (E2E) DAC models which infer dialogue acts directly from audio signals. In this work, we propose an E2E neural architecture that takes into account the need for characterizing prosodic phenomena co-occurring at different levels inside an utterance. A novel part of this architecture is a learnable gating mechanism that assesses the importance of prosodic features and selectively retains core information necessary for E2E DAC. Our proposed model improves DAC accuracy by 1.07% absolute across three publicly available benchmark datasets.

Feng-Ju Chang, Thejaswi Muniyappa, Kanthashree Mysore Sathyendra et al.

Personalization in multi-turn dialogs has been a long standing challenge for end-to-end automatic speech recognition (E2E ASR) models. Recent work on contextual adapters has tackled rare word recognition using user catalogs. This adaptation, however, does not incorporate an important cue, the dialog act, which is available in a multi-turn dialog scenario. In this work, we propose a dialog act guided contextual adapter network. Specifically, it leverages dialog acts to select the most relevant user catalogs and creates queries based on both -- the audio as well as the semantic relationship between the carrier phrase and user catalogs to better guide the contextual biasing. On industrial voice assistant datasets, our model outperforms both the baselines - dialog act encoder-only model, and the contextual adaptation, leading to the most improvement over the no-context model: 58% average relative word error rate reduction (WERR) in the multi-turn dialog scenario, in comparison to the prior-art contextual adapter, which has achieved 39% WERR over the no-context model.

Xuandi Fu, Feng-Ju Chang, Martin Radfar et al.

End-to-end Spoken Language Understanding (E2E SLU) has attracted increasing interest due to its advantages of joint optimization and low latency when compared to traditionally cascaded pipelines. Existing E2E SLU models usually follow a two-stage configuration where an Automatic Speech Recognition (ASR) network first predicts a transcript which is then passed to a Natural Language Understanding (NLU) module through an interface to infer semantic labels, such as intent and slot tags. This design, however, does not consider the NLU posterior while making transcript predictions, nor correct the NLU prediction error immediately by considering the previously predicted word-pieces. In addition, the NLU model in the two-stage system is not streamable, as it must wait for the audio segments to complete processing, which ultimately impacts the latency of the SLU system. In this work, we propose a streamable multi-task semantic transducer model to address these considerations. Our proposed architecture predicts ASR and NLU labels auto-regressively and uses a semantic decoder to ingest both previously predicted word-pieces and slot tags while aggregating them through a fusion network. Using an industry scale SLU and a public FSC dataset, we show the proposed model outperforms the two-stage E2E SLU model for both ASR and NLU metrics.

Min Yang, Cheng Cui, Xuetong Xue et al.

This paper presents the 2nd place solution to the Google Landmark Retrieval Competition 2020. We propose a training method of global feature model for landmark retrieval without post-processing, such as local feature and spatial verification. There are two parts in our retrieval method in this competition. This training scheme mainly includes training by increasing margin value of arcmargin loss and increasing image resolution step by step. Models are trained by PaddlePaddle framework and Pytorch framework, and then converted to tensorflow 2.2. Using this method, we got a public score of 0.40176 and a private score of 0.36278 and achieved 2nd place in the Google Landmark Retrieval Competition 2020.

Yuning Wu, Ke Wang, Devin Chen et al.

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising paradigm for post-training reasoning models. However, group-based methods such as Group Relative Policy Optimization (GRPO) face a critical dilemma in sparse-reward settings: pure Reinforcement Learning (RL) suffers from advantage collapse and high-variance gradient estimation, while mixed-policy optimization introduces persistent distributional bias. To resolve this dilemma, we introduce Hindsight-Anchored Policy Optimization (HAPO). HAPO employs the Synthetic Success Injection (SSI) operator, a hindsight mechanism that selectively anchors optimization to teacher demonstrations during failure. This injection is governed by a Thompson sampling-inspired gating mechanism, creating an autonomous, self-paced curriculum. Theoretically, we demonstrate that HAPO achieves \textit{asymptotic consistency}: by naturally annealing the teacher signal as the policy improves, HAPO recovers the unbiased on-policy gradient. This ensures off-policy guidance acts as a temporary scaffold rather than a persistent ceiling, enabling the model to surpass the limitations of static teacher forcing.

Kai Wei, Raymond Li, Xi Zhu et al.

Retrieval-Augmented Generation (RAG) has emerged as a foundational paradigm for grounding large language models in external knowledge. While adaptive retrieval mechanisms have improved retrieval efficiency, existing approaches treat post-retrieval failure as a signal to retry rather than to diagnose -- leaving the structural causes of query-evidence misalignment unaddressed. We observe that a significant portion of persistent retrieval failures stem not from the absence of relevant evidence but from an alignment gap between the query and the evidence space. We propose Skill-RAG, a failure-aware RAG framework that couples a lightweight hidden-state prober with a prompt-based skill router. The prober gates retrieval at two pipeline stages; upon detecting a failure state, the skill router diagnoses the underlying cause and selects among four retrieval skills -- query rewriting, question decomposition, evidence focusing, and an exit skill for truly irreducible cases -- to correct misalignment before the next generation attempt. Experiments across multiple open-domain QA and complex reasoning benchmarks show that Skill-RAG substantially improves accuracy on hard cases persisting after multi-turn retrieval, with particularly strong gains on out-of-distribution datasets. Representation-space analyses further reveal that the proposed skills occupy structured, separable regions of the failure state space, supporting the view that query-evidence misalignment is a typed rather than monolithic phenomenon.

Yaoning Yu, Kai-Min Chang, Ye Yu et al.

Financial documents like earning reports or balance sheets often involve long tables and multi-page reports. Large language models have become a new tool to help numerical reasoning and understanding these documents. However, prompt quality can have a major effect on how well LLMs perform these financial reasoning tasks. Most current methods tune prompts on fixed datasets of financial text or tabular data, which limits their ability to adapt to new question types or document structures, or they involve costly and manually labeled/curated dataset to help build the prompts. We introduce a self-improving prompt framework driven by data-augmented optimization. In this closed-loop process, we generate synthetic financial tables and document excerpts, verify their correctness and robustness, and then update the prompt based on the results. Specifically, our framework combines a synthetic data generator with verifiers and a prompt optimizer, where the generator produces new examples that exposes weaknesses in the current prompt, the verifiers check the validity and robustness of the produced examples, and the optimizer incrementally refines the prompt in response. By iterating these steps in a feedback cycle, our method steadily improves prompt accuracy on financial reasoning tasks without needing external labels. Evaluation on DocMath-Eval benchmark demonstrates that our system achieves higher performance in both accuracy and robustness than standard prompt methods, underscoring the value of incorporating synthetic data generation into prompt learning for financial applications.

Yiquan Wang, Yahui Ma, Yuhan Chang et al.

Diffusion models have emerged as a leading framework in generative modeling, showing significant potential to accelerate and transform the traditionally slow and costly process of drug discovery. This review provides a systematic comparison of their application in designing two principal therapeutic modalities: small molecules and therapeutic peptides. We analyze how a unified framework of iterative denoising is adapted to the distinct molecular representations, chemical spaces, and design objectives of each modality. For small molecules, these models excel at structure-based design, generating novel, pocket-fitting ligands with desired physicochemical properties, yet face the critical hurdle of ensuring chemical synthesizability. Conversely, for therapeutic peptides, the focus shifts to generating functional sequences and designing de novo structures, where the primary challenges are achieving biological stability against proteolysis, ensuring proper folding, and minimizing immunogenicity. Despite these distinct challenges, both domains face shared hurdles: the need for more accurate scoring functions, the scarcity of high-quality experimental data, and the crucial requirement for experimental validation. We conclude that the full potential of diffusion models will be unlocked by bridging these modality-specific gaps and integrating them into automated, closed-loop Design-Build-Test-Learn (DBTL) platforms, thereby shifting the paradigm from chemical exploration to the targeted creation of novel therapeutics.

Zhaoqian Xue, Guanhong Liu, Kai Wei et al.

Access to health resources is a critical determinant of public well-being and societal resilience, particularly during public health crises when demand for medical services and preventive care surges. However, disparities in accessibility persist across demographic and geographic groups, raising concerns about equity. Traditional survey methods often fall short due to limitations in coverage, cost, and timeliness. This study leverages crowdsourced data from Google Maps reviews, applying advanced natural language processing techniques, specifically ModernBERT, to extract insights on public perceptions of health resource accessibility in the United States during the COVID-19 pandemic. Additionally, we employ Partial Least Squares regression to examine the relationship between accessibility perceptions and key socioeconomic and demographic factors including political affiliation, racial composition, and educational attainment. Our findings reveal that public perceptions of health resource accessibility varied significantly across the U.S., with disparities peaking during the pandemic and slightly easing post-crisis. Political affiliation, racial demographics, and education levels emerged as key factors shaping these perceptions. These findings underscore the need for targeted interventions and policy measures to address inequities, fostering a more inclusive healthcare infrastructure that can better withstand future public health challenges.

Yaoning Yu, Ye Yu, Kai Wei et al.

Prompt quality plays a critical role in the performance of large language models (LLMs), motivating a growing body of work on prompt optimization. Most existing methods optimize prompts over a fixed dataset, assuming static input distributions and offering limited support for iterative improvement. We introduce SIPDO (Self-Improving Prompts through Data-Augmented Optimization), a closed-loop framework for prompt learning that integrates synthetic data generation into the optimization process. SIPDO couples a synthetic data generator with a prompt optimizer, where the generator produces new examples that reveal current prompt weaknesses and the optimizer incrementally refines the prompt in response. This feedback-driven loop enables systematic improvement of prompt performance without assuming access to external supervision or new tasks. Experiments across question answering and reasoning benchmarks show that SIPDO outperforms standard prompt tuning methods, highlighting the value of integrating data synthesis into prompt learning workflows.

Zhijie Duan, Kai Wei, Zhaoqian Xue et al.

Social media is a rich source of real-world data that captures valuable patient experience information for pharmacovigilance. However, mining data from unstructured and noisy social media content remains a challenging task. We present a systematic framework that leverages large language models (LLMs) to extract medication side effects from social media and organize them into a knowledge graph (KG). We apply this framework to semaglutide for weight loss using data from Reddit. Using the constructed knowledge graph, we perform comprehensive analyses to investigate reported side effects across different semaglutide brands over time. These findings are further validated through comparison with adverse events reported in the FAERS database, providing important patient-centered insights into semaglutide's side effects that complement its safety profile and current knowledge base of semaglutide for both healthcare professionals and patients. Our work demonstrates the feasibility of using LLMs to transform social media data into structured KGs for pharmacovigilance.

Jixuan Wang, Martin Radfar, Kai Wei et al.

It is challenging to extract semantic meanings directly from audio signals in spoken language understanding (SLU), due to the lack of textual information. Popular end-to-end (E2E) SLU models utilize sequence-to-sequence automatic speech recognition (ASR) models to extract textual embeddings as input to infer semantics, which, however, require computationally expensive auto-regressive decoding. In this work, we leverage self-supervised acoustic encoders fine-tuned with Connectionist Temporal Classification (CTC) to extract textual embeddings and use joint CTC and SLU losses for utterance-level SLU tasks. Experiments show that our model achieves 4% absolute improvement over the the state-of-the-art (SOTA) dialogue act classification model on the DSTC2 dataset and 1.3% absolute improvement over the SOTA SLU model on the SLURP dataset.

Kai Wei, Thanh Tran, Feng-Ju Chang et al.

Recent years have seen significant advances in end-to-end (E2E) spoken language understanding (SLU) systems, which directly predict intents and slots from spoken audio. While dialogue history has been exploited to improve conventional text-based natural language understanding systems, current E2E SLU approaches have not yet incorporated such critical contextual signals in multi-turn and task-oriented dialogues. In this work, we propose a contextual E2E SLU model architecture that uses a multi-head attention mechanism over encoded previous utterances and dialogue acts (actions taken by the voice assistant) of a multi-turn dialogue. We detail alternative methods to integrate these contexts into the state-ofthe-art recurrent and transformer-based models. When applied to a large de-identified dataset of utterances collected by a voice assistant, our method reduces average word and semantic error rates by 10.8% and 12.6%, respectively. We also present results on a publicly available dataset and show that our method significantly improves performance over a noncontextual baseline

Jixuan Wang, Kai Wei, Martin Radfar et al.

We propose a novel Transformer encoder-based architecture with syntactical knowledge encoded for intent detection and slot filling. Specifically, we encode syntactic knowledge into the Transformer encoder by jointly training it to predict syntactic parse ancestors and part-of-speech of each token via multi-task learning. Our model is based on self-attention and feed-forward layers and does not require external syntactic information to be available at inference time. Experiments show that on two benchmark datasets, our models with only two Transformer encoder layers achieve state-of-the-art results. Compared to the previously best performed model without pre-training, our models achieve absolute F1 score and accuracy improvement of 1.59% and 0.85% for slot filling and intent detection on the SNIPS dataset, respectively. Our models also achieve absolute F1 score and accuracy improvement of 0.1% and 0.34% for slot filling and intent detection on the ATIS dataset, respectively, over the previously best performed model. Furthermore, the visualization of the self-attention weights illustrates the benefits of incorporating syntactic information during training.

Cheng Cui, Zhi Ye, Yangxi Li et al.

Simi-Supervised Recognition Challenge-FGVC7 is a challenging fine-grained recognition competition. One of the difficulties of this competition is how to use unlabeled data. We adopted pseudo-tag data mining to increase the amount of training data. The other one is how to identify similar birds with a very small difference, especially those have a relatively tiny main-body in examples. We combined generic image recognition and fine-grained image recognition method to solve the problem. All generic image recognition models were training using PaddleClas . Using the combination of two different ways of deep recognition models, we finally won the third place in the competition.

Rishabh Iyer, John T. Halloran, Kai Wei

This paper introduces Jensen, an easily extensible and scalable toolkit for production-level machine learning and convex optimization. Jensen implements a framework of convex (or loss) functions, convex optimization algorithms (including Gradient Descent, L-BFGS, Stochastic Gradient Descent, Conjugate Gradient, etc.), and a family of machine learning classifiers and regressors (Logistic Regression, SVMs, Least Square Regression, etc.). This framework makes it possible to deploy and train models with a few lines of code, and also extend and build upon this by integrating new loss functions and optimization algorithms.

John Moore, Joel Pfeiffer, Kai Wei et al.

In real world systems, the predictions of deployed Machine Learned models affect the training data available to build subsequent models. This introduces a bias in the training data that needs to be addressed. Existing solutions to this problem attempt to resolve the problem by either casting this in the reinforcement learning framework or by quantifying the bias and re-weighting the loss functions. In this work, we develop a novel Adversarial Neural Network (ANN) model, an alternative approach which creates a representation of the data that is invariant to the bias. We take the Paid Search auction as our working example and ad display position features as the confounding features for this setting. We show the success of this approach empirically on both synthetic data as well as real world paid search auction data from a major search engine.

Kai Wei, Rishabh Iyer, Shengjie Wang et al.

We study two mixed robust/average-case submodular partitioning problems that we collectively call Submodular Partitioning. These problems generalize both purely robust instances of the problem (namely max-min submodular fair allocation (SFA) and min-max submodular load balancing (SLB) and also generalize average-case instances (that is the submodular welfare problem (SWP) and submodular multiway partition (SMP). While the robust versions have been studied in the theory community, existing work has focused on tight approximation guarantees, and the resultant algorithms are not, in general, scalable to very large real-world applications. This is in contrast to the average case, where most of the algorithms are scalable. In the present paper, we bridge this gap, by proposing several new algorithms (including those based on greedy, majorization-minimization, minorization-maximization, and relaxation algorithms) that not only scale to large sizes but that also achieve theoretical approximation guarantees close to the state-of-the-art, and in some cases achieve new tight bounds. We also provide new scalable algorithms that apply to additive combinations of the robust and average-case extreme objectives. We show that these problems have many applications in machine learning (ML). This includes: 1) data partitioning and load balancing for distributed machine algorithms on parallel machines; 2) data clustering; and 3) multi-label image segmentation with (only) Boolean submodular functions via pixel partitioning. We empirically demonstrate the efficacy of our algorithms on real-world problems involving data partitioning for distributed optimization of standard machine learning objectives (including both convex and deep neural network objectives), and also on purely unsupervised (i.e., no supervised or semi-supervised learning, and no interactive segmentation) image segmentation.