Cunli Mao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Cunli Mao has a total of 62 co-authors. Here are the top co-authors by number of shared publications:
- Zhengtao Yu
Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming, China
15 shared publications - Shengxiang Gao
Kunming University of Science and Technology, Kunming, China
7 shared publications - Yuxin Huang
School of Information Engineering and Automation, Kunming University of Science and Technology, Kunming, China
7 shared publications - Linqin Wang
Kunming University of Science and Technology, Kunming, China
4 shared publications - Shengxiang Gao
Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming China
4 shared publications - Ran Song
Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming, China
3 shared publications
Latest Publications
- Open Access
- 8 September 2026
- Published by Elsevier in SSRN Electronic Journal
Probabilistic Short-Term Electricity Price Forecasting Considering Renewable Energy Penetration and Residual Load
High penetration of renewable energy increases the nonlinearity, state dependence, and uncertainty of electricity-market supply–demand relations and price formation. To improve both short-term electricity-price forecasting accuracy and its usefulness for downstream decisions, this study develops a probabilistic forecasting and energy-storage evaluation framework centered on renewable energy penetration and residual load. First, market-state variables are constructed from load, wind and solar generation, and generation status, and their relationships with electricity prices are examined across 38 European bidding zones. Second, a one-hour-ahead point forecasting model is developed using least-squares boosting (LSBoost), and fixed, state-conditioned, and online rolling prediction intervals are constructed using conformal calibration. Finally, the point and interval forecasts are fed into a unified storage charging/discharging rule to evaluate operating profit and risk. The results show that residual load is positively aligned with price in all 38 valid markets, while renewable penetration is negatively aligned with price in all 37 markets with valid penetration data. In DE-LU, LSBoost achieves an MAE of 8.019 EUR/MWh, 27.61% lower than Ridge. Online rolling asymmetric calibration increases the empirical coverage of the nominal 80% prediction interval from 71.88% to 79.67%. In the storage evaluation, the LSBoost point-forecast strategy yields the highest total profit of 53.35 million EUR, whereas the rolling interval-boundary strategy reduces maximum drawdown from 0.388 to 0.363 million EUR and equivalent full cycles from 59.22 to 50.03. These results demonstrate that point-forecast accuracy, interval performance, and storage-operation outcomes are not fully aligned, and that interval forecasts provide complementary information for operational risk control.
- 8 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessProbabilistic Short-Term Electricity Price Forecasting Considering Renewable Energy Penetration and Residual Load
High penetration of renewable energy increases the nonlinearity, state dependence, and uncertainty of electricity-market supply–demand relations and price formation. To improve both short-term electricity-price forecasting accuracy and its usefulness for downstream decisions, this study develops a probabilistic forecasting and energy-storage evaluation framework centered on renewable energy penetration and residual load. First, market-state variables are constructed from load, wind and solar generation, and generation status, and their relationships with electricity prices are examined across 38 European bidding zones. Second, a one-hour-ahead point forecasting model is developed using least-squares boosting (LSBoost), and fixed, state-conditioned, and online rolling prediction intervals are constructed using conformal calibration. Finally, the point and interval forecasts are fed into a unified storage charging/discharging rule to evaluate operating profit and risk. The results show that residual load is positively aligned with price in all 38 valid markets, while renewable penetration is negatively aligned with price in all 37 markets with valid penetration data. In DE-LU, LSBoost achieves an MAE of 8.019 EUR/MWh, 27.61% lower than Ridge. Online rolling asymmetric calibration increases the empirical coverage of the nominal 80% prediction interval from 71.88% to 79.67%. In the storage evaluation, the LSBoost point-forecast strategy yields the highest total profit of 53.35 million EUR, whereas the rolling interval-boundary strategy reduces maximum drawdown from 0.388 to 0.363 million EUR and equivalent full cycles from 59.22 to 50.03. These results demonstrate that point-forecast accuracy, interval performance, and storage-operation outcomes are not fully aligned, and that interval forecasts provide complementary information for operational risk control.
- Open Access
- 5 September 2026
- Published by Elsevier in SSRN Electronic Journal
Probabilistic Short-Term Electricity Price Forecasting Considering Renewable Energy Penetration and Residual Load
High penetration of renewable energy increases the nonlinearity, state dependence, and uncertainty of electricity-market supply–demand relations and price formation. To improve both short-term electricity-price forecasting accuracy and its usefulness for downstream decisions, this study develops a probabilistic forecasting and energy-storage evaluation framework centered on renewable energy penetration and residual load. First, market-state variables are constructed from load, wind and solar generation, and generation status, and their relationships with electricity prices are examined across 38 European bidding zones. Second, a one-hour-ahead point forecasting model is developed using least-squares boosting (LSBoost), and fixed, state-conditioned, and online rolling prediction intervals are constructed using conformal calibration. Finally, the point and interval forecasts are fed into a unified storage charging/discharging rule to evaluate operating profit and risk. The results show that residual load is positively aligned with price in all 38 valid markets, while renewable penetration is negatively aligned with price in all 37 markets with valid penetration data. In DE-LU, LSBoost achieves an MAE of 8.019 EUR/MWh, 27.61% lower than Ridge. Online rolling asymmetric calibration increases the empirical coverage of the nominal 80% prediction interval from 71.88% to 79.67%. In the storage evaluation, the LSBoost point-forecast strategy yields the highest total profit of 53.35 million EUR, whereas the rolling interval-boundary strategy reduces maximum drawdown from 0.388 to 0.363 million EUR and equivalent full cycles from 59.22 to 50.03. These results demonstrate that point-forecast accuracy, interval performance, and storage-operation outcomes are not fully aligned, and that interval forecasts provide complementary information for operational risk control.
- 5 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessProbabilistic Short-Term Electricity Price Forecasting Considering Renewable Energy Penetration and Residual Load
High penetration of renewable energy increases the nonlinearity, state dependence, and uncertainty of electricity-market supply–demand relations and price formation. To improve both short-term electricity-price forecasting accuracy and its usefulness for downstream decisions, this study develops a probabilistic forecasting and energy-storage evaluation framework centered on renewable energy penetration and residual load. First, market-state variables are constructed from load, wind and solar generation, and generation status, and their relationships with electricity prices are examined across 38 European bidding zones. Second, a one-hour-ahead point forecasting model is developed using least-squares boosting (LSBoost), and fixed, state-conditioned, and online rolling prediction intervals are constructed using conformal calibration. Finally, the point and interval forecasts are fed into a unified storage charging/discharging rule to evaluate operating profit and risk. The results show that residual load is positively aligned with price in all 38 valid markets, while renewable penetration is negatively aligned with price in all 37 markets with valid penetration data. In DE-LU, LSBoost achieves an MAE of 8.019 EUR/MWh, 27.61% lower than Ridge. Online rolling asymmetric calibration increases the empirical coverage of the nominal 80% prediction interval from 71.88% to 79.67%. In the storage evaluation, the LSBoost point-forecast strategy yields the highest total profit of 53.35 million EUR, whereas the rolling interval-boundary strategy reduces maximum drawdown from 0.388 to 0.363 million EUR and equivalent full cycles from 59.22 to 50.03. These results demonstrate that point-forecast accuracy, interval performance, and storage-operation outcomes are not fully aligned, and that interval forecasts provide complementary information for operational risk control.
- Open Access
- 13 July 2026
- Published by Elsevier in SSRN Electronic Journal
Enhancing Multilingual Dependency Parsing via Dynamic Linguistic Feature Extraction and Alignment
Multi-lingual dependency parsing aims to extract useful syntactic knowledge from multiple high-resource languages to improve the parsing performance of low-resource target languages. How to effectively extract linguistic commonality structures and reduce divergence feature interferences is still a key challenge. To alleviate these issues, we propose a dynamic linguistic feature extraction and alignment framework for multi-lingual dependency parsing. On the one hand, we construct multi-lingual parallel dependency trees through confidence-aware traditional parser and LLM collaboration, thus explicitly extracting cross-lingual aligned syntactic knowledge. On the other hand, we introduce a dynamic feature alignment network to selectively transfer useful high-resource syntactic features and suppress harmful ones.Experiments on four low-resource benchmarks demonstrate that our model consistently improves multi-lingual parsing baselines, achieving competitive or superior performance on all languages. Further analyses indicate that the proposed framework effectively captures transferable structural knowledge and reduces negative transfer, leading to better parsing performance in low-resource settings. Our codes and data are available at \url{https://github.com/**}
- 13 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEnhancing Multilingual Dependency Parsing via Dynamic Linguistic Feature Extraction and Alignment
Multi-lingual dependency parsing aims to extract useful syntactic knowledge from multiple high-resource languages to improve the parsing performance of low-resource target languages. How to effectively extract linguistic commonality structures and reduce divergence feature interferences is still a key challenge. To alleviate these issues, we propose a dynamic linguistic feature extraction and alignment framework for multi-lingual dependency parsing. On the one hand, we construct multi-lingual parallel dependency trees through confidence-aware traditional parser and LLM collaboration, thus explicitly extracting cross-lingual aligned syntactic knowledge. On the other hand, we introduce a dynamic feature alignment network to selectively transfer useful high-resource syntactic features and suppress harmful ones.Experiments on four low-resource benchmarks demonstrate that our model consistently improves multi-lingual parsing baselines, achieving competitive or superior performance on all languages. Further analyses indicate that the proposed framework effectively captures transferable structural knowledge and reduces negative transfer, leading to better parsing performance in low-resource settings. Our codes and data are available at \url{https://github.com/**}
- 28 February 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (4) , 152
Personalized federated multilingual machine translation with multi-level knowledge fusion
- 28 February 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (4) , 152
Personalized federated multilingual machine translation with multi-level knowledge fusion
- 23 November 2025
- Published by Springer Nature
- p. 88-100
A Southeast Asian Language OCR Dataset and Evaluation for Large Multimodal Models
- 23 November 2025
- Published by Springer Nature
- p. 88-100
A Southeast Asian Language OCR Dataset and Evaluation for Large Multimodal Models
- 23 November 2025
- Published by Springer Nature
- p. 115-127
Fine-Grained Prosody-Controllable Lao Speech Synthesis Guided by Natural Language
- 23 November 2025
- Published by Springer Nature
- p. 115-127
Fine-Grained Prosody-Controllable Lao Speech Synthesis Guided by Natural Language
- Open Access
Multilingual Knowledge Graph Completion via Efficient Multilingual Knowledge Sharing
Large language models (LLMs) based Multilingual Knowledge Graph Completion (MKGC) aim to predict missing facts by leveraging LLMs' multilingual understanding capabilities, improving the completeness of multilingual knowledge graphs (KGs). However, existing MKGC research underutilizes the multilingual capabilities of LLMs and ignores the shareability of cross-lingual knowledge. In this paper, we propose a novel MKGC framework that leverages multilingual shared knowledge to significantly enhance performance through two components: Knowledge-level Grouped Mixture of Experts (KL-GMoE) and Iterative Entity Reranking (IER). KL-GMoE efficiently models shared knowledge, while IER significantly enhances its utilization. To evaluate our framework, we constructed a mKG dataset containing 5 languages and conducted comprehensive comparative experiments with existing state-of-the-art (SOTA) MKGC method. The experimental results demonstrate that our framework achieves improvements of 5.47%, 3.27%, and 1.01% in the Hits@1, Hits@3, and Hits@10 metrics, respectively, compared with SOTA MKGC method. Further experimental analysis revealed the properties of knowledge sharing in settings of unseen and unbalanced languages. We have released the dataset and code for our work on https://github.com/gaoxiaofei07/KL-GMoE.
Open AccessMultilingual Knowledge Graph Completion via Efficient Multilingual Knowledge Sharing
Large language models (LLMs) based Multilingual Knowledge Graph Completion (MKGC) aim to predict missing facts by leveraging LLMs' multilingual understanding capabilities, improving the completeness of multilingual knowledge graphs (KGs). However, existing MKGC research underutilizes the multilingual capabilities of LLMs and ignores the shareability of cross-lingual knowledge. In this paper, we propose a novel MKGC framework that leverages multilingual shared knowledge to significantly enhance performance through two components: Knowledge-level Grouped Mixture of Experts (KL-GMoE) and Iterative Entity Reranking (IER). KL-GMoE efficiently models shared knowledge, while IER significantly enhances its utilization. To evaluate our framework, we constructed a mKG dataset containing 5 languages and conducted comprehensive comparative experiments with existing state-of-the-art (SOTA) MKGC method. The experimental results demonstrate that our framework achieves improvements of 5.47%, 3.27%, and 1.01% in the Hits@1, Hits@3, and Hits@10 metrics, respectively, compared with SOTA MKGC method. Further experimental analysis revealed the properties of knowledge sharing in settings of unseen and unbalanced languages. We have released the dataset and code for our work on https://github.com/gaoxiaofei07/KL-GMoE.
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (24) , 25380-25388
SECodec: Structural Entropy-based Compressive Speech Representation Codec for Speech Language Models
With the rapid advancement of large language models (LLMs), discrete speech representations have become crucial for integrating speech into LLMs. Existing methods for speech representation discretization rely on a predefined codebook size and Euclidean distance-based quantization. However, 1) the size of codebook is a critical parameter that affects both codec performance and downstream task training efficiency. 2) The Euclidean distance-based quantization may lead to audio distortion when the size of the codebook is controlled within a reasonable range. In fact, in the field of information compression, structural information and entropy guidance are crucial, but previous methods have largely overlooked these factors. Therefore, we address the above issues from an information-theoretic perspective, we present SECodec, a novel speech representation codec based on structural entropy (SE) for building speech language models. Specifically, we first model speech as a graph, clustering the speech features nodes within the graph and extracting the corresponding codebook by hierarchically and disentangledly minimizing 2D SE. Then, to address the issue of audio distortion, we propose a new quantization method. This method still adheres to the 2D SE minimization principle, adaptively selecting the most suitable token corresponding to the cluster for each incoming original speech node. Furthermore, we develop a Structural Entropy-based Speech Language Model (SESLM) that leverages SECodec. Experimental results demonstrate that SECodec performs comparably to EnCodec in speech reconstruction, and SESLM surpasses VALL-E in zero-shot text-to-speech tasks.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (24) , 25380-25388
Open AccessSECodec: Structural Entropy-based Compressive Speech Representation Codec for Speech Language Models
With the rapid advancement of large language models (LLMs), discrete speech representations have become crucial for integrating speech into LLMs. Existing methods for speech representation discretization rely on a predefined codebook size and Euclidean distance-based quantization. However, 1) the size of codebook is a critical parameter that affects both codec performance and downstream task training efficiency. 2) The Euclidean distance-based quantization may lead to audio distortion when the size of the codebook is controlled within a reasonable range. In fact, in the field of information compression, structural information and entropy guidance are crucial, but previous methods have largely overlooked these factors. Therefore, we address the above issues from an information-theoretic perspective, we present SECodec, a novel speech representation codec based on structural entropy (SE) for building speech language models. Specifically, we first model speech as a graph, clustering the speech features nodes within the graph and extracting the corresponding codebook by hierarchically and disentangledly minimizing 2D SE. Then, to address the issue of audio distortion, we propose a new quantization method. This method still adheres to the 2D SE minimization principle, adaptively selecting the most suitable token corresponding to the cluster for each incoming original speech node. Furthermore, we develop a Structural Entropy-based Speech Language Model (SESLM) that leverages SECodec. Experimental results demonstrate that SECodec performs comparably to EnCodec in speech reconstruction, and SESLM surpasses VALL-E in zero-shot text-to-speech tasks.
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (1) , 541-549
BiDeV: Bilateral Defusing Verification for Complex Claim Fact-Checking
Complex claim fact-checking performs a crucial role in disinformation detection. However, existing fact-checking methods struggle with claim vagueness, specifically in effectively handling latent information and complex relations within claims. Moreover, evidence redundancy, where non-essential information complicates the verification process, remains a significant issue. To tackle these limitations, we propose Bilateral Defusing Verification (BiDeV), a novel fact-checking working-flow framework integrating multiple role-played LLMs to mimic the human-expert fact-checking process. BiDeV consists of two main modules: Vagueness Defusing identifies latent information and resolves complex relations to simplify the claim, and Redundancy Defusing eliminates redundant content to enhance the evidence quality. Extensive experimental results on two widely used challenging fact-checking benchmarks (Hover and Feverous-s) demonstrate that our BiDeV can achieve the best performance under both gold and open settings. This highlights the effectiveness of BiDeV in handling complex claims and ensuring precise fact-checking.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (1) , 541-549
Open AccessBiDeV: Bilateral Defusing Verification for Complex Claim Fact-Checking
Complex claim fact-checking performs a crucial role in disinformation detection. However, existing fact-checking methods struggle with claim vagueness, specifically in effectively handling latent information and complex relations within claims. Moreover, evidence redundancy, where non-essential information complicates the verification process, remains a significant issue. To tackle these limitations, we propose Bilateral Defusing Verification (BiDeV), a novel fact-checking working-flow framework integrating multiple role-played LLMs to mimic the human-expert fact-checking process. BiDeV consists of two main modules: Vagueness Defusing identifies latent information and resolves complex relations to simplify the claim, and Redundancy Defusing eliminates redundant content to enhance the evidence quality. Extensive experimental results on two widely used challenging fact-checking benchmarks (Hover and Feverous-s) demonstrate that our BiDeV can achieve the best performance under both gold and open settings. This highlights the effectiveness of BiDeV in handling complex claims and ensuring precise fact-checking.
- 24 February 2025
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- p. 1-11
Personalized Lao language synthesis via disentangled neural codec language model
- 24 February 2025
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- p. 1-11
Personalized Lao language synthesis via disentangled neural codec language model