❌

Normal view

Received — 15 September 2026 ⏭

Label-Guided Knowledge Distillation for 3D-CNNs in Action Recognition

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yanjiang Shi · Peng Zhao · Nan Qi · Guiqin Wang
14 September 2026 at 12:00
arXiv:2609.13024v1 Announce Type: cross Abstract: As a key model compression technique, knowledge distillation aims to transfer knowledge from a high-capacity teacher model to a lightweight student model for enhancing the latter's performance. In this work, we reviewed the feature knowledge distillation for 3D-CNNs and observed that most feature distillation methods in video analysis are simple adaptations of those used in image analysis, often neglecting the differences of video features in the temporal dimension. To address this issue, we proposed Label-Guided Knowledge Distillation (LGKD) to guide the distillation of student model features using ground truth labels. Our method entails two components: sample-wise distillation and class-wise distillation, enabling the student model to learn feature representation of the teacher model at two levels. Sample-wise distillation utilizes label information and the teacher's probability distribution to guide the learning of features that significantly impact temporal accuracy while mitigating noise. Meanwhile, class-wise feature distillation employs a prototype network to further capture the relational knowledge among samples within the same category, enhancing the student's ability to learn higher-dimensional semantic information and improving model generalization. To demonstrate the effectiveness and superiority of our method, we conducted comprehensive experiments on two benchmark action recognition datasets, UCF101 and HMDB51, achieving competitive results.
Received — 27 May 2026 ⏭

Raspberry aqueous extract ameliorates MAFLD in mice by regulating gut microbiota and purine metabolism

✇Omics in Hepatocellular
By: Yanyan Gao · Nian Liu · Huaxin Wei · Tianchi Sun · Fangying Xu · Lin Chen · Jiannan Qiu · Xiaobing Dou
18 May 2026 at 18:00

Front Nutr. 2026 Apr 30;13:1818086. doi: 10.3389/fnut.2026.1818086. eCollection 2026.

ABSTRACT

INTRODUCTION: Metabolism-associated fatty liver disease (MAFLD) has emerged as a severe worldwide public health burden with insufficient available clinical therapeutic strategies, which underscores the urgent demand for safe, natural dietary interventions. Raspberry (Rubus idaeus L.), a typical food-medicine homologous fruit abundant in diverse bioactive components including anthocyanins, flavonoids and polysaccharides, possesses prominent nutritional and medicinal potential.

METHODS: In this study, raspberry aqueous extract (RE) was prepared to comprehensively investigate its ameliorative effects and underlying molecular mechanisms against MAFLD. MAFLD animal model was established in C57BL/6 mice via 12-week high-fat diet (HFD) feeding. From the 9th week, model mice were intragastrically administered with RE at doses of 1 g/kg/d and 2 g/kg/d for continuous intervention. Integrated multi-omics analyses including 16S rRNA microbial sequencing, serum/hepatic biochemical detection, histopathological examination, in vivo microbial colonization assay, and in vitro cellular and metabolomic experiments were performed to systematically clarify the regulatory mechanism.

RESULTS: RE treatment markedly improved the core pathological phenotypes of MAFLD mice, and significantly mitigated hepatic steatosis and hepatocellular injury. 16S rRNA sequencing demonstrated that RE remodeled the gut microbial dysbiosis, specifically elevating the abundance of beneficial genus Ileibacterium and suppressing pathogenic microbial taxa. Meanwhile, RE strengthened intestinal mucosal barrier integrity by upregulating tight junction protein expression, and activated hepatic purine metabolic reprogramming to boost the levels of critical metabolites including inosine and ADP. Spearman correlation analysis verified the significantly positive correlation between Ileibacterium abundance and hepatic inosine content, and both factors were closely correlated with the remission of MAFLD pathological indicators. In vivo colonization experiments further validated that Ileibacterium intervention alone remarkably alleviated hepatic lipid deposition and liver damage in MAFLD mice. In vitro strain metabolomics confirmed that Ileibacterium could directly biosynthesize and secrete inosine extracellularly. Furthermore, in vitro AML12 hepatocyte experiments revealed that 100 μM inosine remarkably relieved palmitic acid-induced lipotoxicity via reducing intracellular lipid overload, reactive oxygen species (ROS) accumulation and mitochondrial dysfunction, alongside modulating the expression of lipid metabolism, inflammatory and autophagy-related genes.

DISCUSSION: Collectively, our results elucidate that raspberry aqueous extract alleviates experimental MAFLD through the gut microbiota-purine metabolism-inosine regulatory axis, in which Ileibacterium and inosine act as the core synergistic mediators. This study provides solid preclinical experimental evidence for the development and application of raspberry as a promising functional food for the prevention and nutritional intervention of MAFLD.

PMID:42146077 | PMC:PMC13171365 | DOI:10.3389/fnut.2026.1818086

Received — 8 April 2026 ⏭

Reflection of Episodes: Learning to Play Game from Expert and Self Experiences

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiaojie Xu · Zongyuan Li · Chang Lu · Runnan Qi · Yanan Ni · Lumin Jiang · Xiangbei Liu · Xuebo Zhang · Yongchun Fang · Kuihua Huang · Xian Guo · Zhanghua Wu · Zhenya Li
7 April 2026 at 12:00
arXiv:2502.13388v3 Announce Type: replace Abstract: StarCraft II is a complex and dynamic real-time strategy (RTS) game environment, which is very suitable for artificial intelligence and reinforcement learning research. To address the problem of Large Language Model(LLM) learning in complex environments through self-reflection, we propose a Reflection of Episodes(ROE) framework based on expert experience and self-experience. This framework first obtains key information in the game through a keyframe selection method, then makes decisions based on expert experience and self-experience. After a game is completed, it reflects on the previous experience to obtain new self-experience. Finally, in the experiment, our method beat the robot under the Very Hard difficulty in TextStarCraft II. We analyze the data of the LLM in the process of the game in detail, verified its effectiveness.
Received — 4 April 2026 ⏭

Reflection of Episodes: Learning to Play Game from Expert and Self Experiences

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiaojie Xu · Zongyuan Li · Chang Lu · Runnan Qi · Yanan Ni · Lumin Jiang · Xiangbei Liu · Xuebo Zhang · Yongchun Fang · Kuihua Huang · Xian Guo · Zhanghua Wu · Zhenya Li
3 April 2026 at 12:00
arXiv:2502.13388v2 Announce Type: replace Abstract: StarCraft II is a complex and dynamic real-time strategy (RTS) game environment, which is very suitable for artificial intelligence and reinforcement learning research. To address the problem of Large Language Model(LLM) learning in complex environments through self-reflection, we propose a Reflection of Episodes(ROE) framework based on expert experience and self-experience. This framework first obtains key information in the game through a keyframe selection method, then makes decisions based on expert experience and self-experience. After a game is completed, it reflects on the previous experience to obtain new self-experience. Finally, in the experiment, our method beat the robot under the Very Hard difficulty in TextStarCraft II. We analyze the data of the LLM in the process of the game in detail, verified its effectiveness.
Received — 17 February 2026 ⏭

A Disentangled Representation Learning Framework for Low-altitude Network Coverage Prediction

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiaojie Li · Zhijie Cai · Nan Qi · Chao Dong · Guangxu Zhu · Haixia Ma · Qihui Wu · Shi Jin
17 February 2026 at 13:00
arXiv:2507.14186v2 Announce Type: replace-cross Abstract: The expansion of the low-altitude economy has underscored the significance of Low-Altitude Network Coverage (LANC) prediction for designing aerial corridors. While accurate LANC forecasting hinges on the antenna beam patterns of Base Stations (BSs), these patterns are typically proprietary and not readily accessible. Operational parameters of BSs, which inherently contain beam information, offer an opportunity for data-driven low-altitude coverage prediction. However, collecting extensive low-altitude road test data is cost-prohibitive, often yielding only sparse samples per BS. This scarcity results in two primary challenges: imbalanced feature sampling due to limited variability in high-dimensional operational parameters against the backdrop of substantial changes in low-dimensional sampling locations, and diminished generalizability stemming from insufficient data samples. To overcome these obstacles, we introduce a dual strategy comprising expert knowledge-based feature compression and disentangled representation learning. The former reduces feature space complexity by leveraging communications expertise, while the latter enhances model generalizability through the integration of propagation models and distinct subnetworks that capture and aggregate the semantic representations of latent features. Experimental evaluation confirms the efficacy of our framework, yielding a 7% reduction in error compared to the best baseline algorithm. Real-network validations further attest to its reliability, achieving practical prediction accuracy with MAE errors at the 5dB level.
There are no more articles
❌