Self-Consistent Adjoint Policy Iteration for Constrained Dynamic Portfolio Choice
研究通过自洽伴随策略迭代方法解决了带约束的动态投资组合选择问题，适用于可预测回报和凸约束情况。
0 citationsRead paper
Selected work
Representative Papers
Learning from Multimodal Pseudo-Labels for Robust Open-Vocabulary Instance and Panoptic Segmentation
This work addresses key challenges in open-vocabulary instance segmentation and open-set panoptic segmentation, including noisy pseudo-labels, weak vision–language alignment, and difficulties in handling out-of-vocabulary categories. To tackle these issues, the authors propose a multimodal pseudo-labeling and training framework that integrates pretrained models such as Grounded SAM, LLaVA, and CLIP. The approach employs a target-vocabulary-guided pseudo-labeling mechanism, CLIP-driven synonym filtering, and GPT-enhanced caption reconstruction to construct semantically consistent vision–text pairs. By jointly optimizing an extended visual grounding loss, a semantic consistency loss, and a generative caption reconstruction loss, the model achieves significantly improved generalization to unseen categories. Evaluated on the COCO benchmark, the method sets new state-of-the-art results in both tasks.
0 citationsRead paper
TELLME: Test-Enhanced Learning for Language Model Enrichment
This work addresses the challenges of large-scale data acquisition and high computational costs in domain adaptation through continual pretraining by introducing, for the first time, Test-Enhanced Learning (TEL) into the continual pretraining framework. By integrating an embedded quiz mechanism, the proposed approach enhances the model’s efficiency in acquiring domain-specific knowledge and its ability to retain long-term memory. Empirical results demonstrate that this method significantly improves domain adaptation efficiency, achieving up to a 23.6% performance gain on financial-domain tasks and a 9.8% improvement in long-term memory retention. The study thus establishes a novel paradigm for effective and cost-efficient domain adaptation.
0 citationsRead paper
Relevance-Aware Rule: Structural Deletion of Irrelevant Conditions in Decision Trees
Decision trees may introduce irrelevant conditions (IRCs)—splitting criteria inconsistent with the class labels of their descendant leaf nodes—due to their binary partitioning mechanism, thereby compromising rule simplicity and reliability. This work is the first to elucidate the structural origins of IRCs and proposes a rigorous diagnostic and pruning framework. By analyzing inverse shifts in class proportions between parent and child nodes, and integrating structural linkage matching, directional consistency checks, and predictive reliability assessment, the method precisely identifies and removes conditions that are both structurally and empirically irrelevant while preserving predictive performance. The resulting rules are substantially simplified without sacrificing accuracy, achieving an effective balance between interpretability and fidelity to the original model’s predictions.
0 citationsRead paper
Not All Prediction Targets Keep Training-Free Diffusion Guidance on the Manifold
This work addresses a critical limitation of Training-Free Guidance (TFG) in high-noise regimes, where estimating a clean image from pure noise often causes the guidance signal to deviate from the data manifold, degrading generation quality. The study is the first to reveal that the choice of prediction target—ε, v, or x—profoundly influences TFG’s ability to preserve the data manifold. Theoretical analysis demonstrates that x-prediction directly yields the clean image, substantially reducing estimation error under high noise. To detect such manifold distortions overlooked by conventional metrics, the authors introduce a novel evaluation measure, guided-class FID (Child FID). Experiments on a newly curated fine-grained bird benchmark and style transfer tasks confirm that TFG with x-prediction significantly outperforms other prediction strategies in maintaining sample manifold consistency.
0 citationsRead paper
Recent publications
Latest Papers
Self-Consistent Adjoint Policy Iteration for Constrained Dynamic Portfolio Choice
研究通过自洽伴随策略迭代方法解决了带约束的动态投资组合选择问题，适用于可预测回报和凸约束情况。
0 citationsRead paper
Learning from Multimodal Pseudo-Labels for Robust Open-Vocabulary Instance and Panoptic Segmentation
This work addresses key challenges in open-vocabulary instance segmentation and open-set panoptic segmentation, including noisy pseudo-labels, weak vision–language alignment, and difficulties in handling out-of-vocabulary categories. To tackle these issues, the authors propose a multimodal pseudo-labeling and training framework that integrates pretrained models such as Grounded SAM, LLaVA, and CLIP. The approach employs a target-vocabulary-guided pseudo-labeling mechanism, CLIP-driven synonym filtering, and GPT-enhanced caption reconstruction to construct semantically consistent vision–text pairs. By jointly optimizing an extended visual grounding loss, a semantic consistency loss, and a generative caption reconstruction loss, the model achieves significantly improved generalization to unseen categories. Evaluated on the COCO benchmark, the method sets new state-of-the-art results in both tasks.
0 citationsRead paper
TELLME: Test-Enhanced Learning for Language Model Enrichment
This work addresses the challenges of large-scale data acquisition and high computational costs in domain adaptation through continual pretraining by introducing, for the first time, Test-Enhanced Learning (TEL) into the continual pretraining framework. By integrating an embedded quiz mechanism, the proposed approach enhances the model’s efficiency in acquiring domain-specific knowledge and its ability to retain long-term memory. Empirical results demonstrate that this method significantly improves domain adaptation efficiency, achieving up to a 23.6% performance gain on financial-domain tasks and a 9.8% improvement in long-term memory retention. The study thus establishes a novel paradigm for effective and cost-efficient domain adaptation.
0 citationsRead paper
Relevance-Aware Rule: Structural Deletion of Irrelevant Conditions in Decision Trees
Decision trees may introduce irrelevant conditions (IRCs)—splitting criteria inconsistent with the class labels of their descendant leaf nodes—due to their binary partitioning mechanism, thereby compromising rule simplicity and reliability. This work is the first to elucidate the structural origins of IRCs and proposes a rigorous diagnostic and pruning framework. By analyzing inverse shifts in class proportions between parent and child nodes, and integrating structural linkage matching, directional consistency checks, and predictive reliability assessment, the method precisely identifies and removes conditions that are both structurally and empirically irrelevant while preserving predictive performance. The resulting rules are substantially simplified without sacrificing accuracy, achieving an effective balance between interpretability and fidelity to the original model’s predictions.
0 citationsRead paper
Not All Prediction Targets Keep Training-Free Diffusion Guidance on the Manifold
This work addresses a critical limitation of Training-Free Guidance (TFG) in high-noise regimes, where estimating a clean image from pure noise often causes the guidance signal to deviate from the data manifold, degrading generation quality. The study is the first to reveal that the choice of prediction target—ε, v, or x—profoundly influences TFG’s ability to preserve the data manifold. Theoretical analysis demonstrates that x-prediction directly yields the clean image, substantially reducing estimation error under high noise. To detect such manifold distortions overlooked by conventional metrics, the authors introduce a novel evaluation measure, guided-class FID (Child FID). Experiments on a newly curated fine-grained bird benchmark and style transfer tasks confirm that TFG with x-prediction significantly outperforms other prediction strategies in maintaining sample manifold consistency.
0 citationsRead paper