Recent years have seen a rapid surge in research leveraging Large Language Models (LLMs) for recommendation. These methods typically employ supervised fine-tuning (SFT) to adapt LLMs to recommendation scenarios, and utilize beam search during inference to efficiently retrieve B top-ranked recommended items. However, we identify a critical training-inference inconsistency: while SFT optimizes the overall probability of positive items, it does not guarantee that such items will be retrieved by beam search even if they possess high overall probabilities. Due to the greedy pruning mechanism, beam search can prematurely discard a positive item once its prefix probability is insufficient.

Loss functions play a pivotal role in optimizing recommendation models. Among various loss functions, Softmax Loss (SL) and Cosine Contrastive Loss (CCL) are particularly effective. Their theoretical connections and differences warrant in-depth exploration. This work conducts comprehensive analyses of these losses, yielding significant insights: 1) Common strengths --- both can be viewed as augmentations of traditional losses with Distributional Robust Optimization (DRO), enhancing robustness to distributional shifts; 2) Respective limitations --- stemming from their use of different distribution distance metrics in DRO optimization, SL exhibits high sensitivity to false negative instances, whereas CCL suffers from low data utilization. To address these limitations, this work proposes a new loss function, DrRL, which generalizes SL and CCL by leveraging Rényi-divergence in DRO optimization. DrRL incorporates the advantageous structures of both SL and CCL, and can be demonstrated to effectively mitigate their limitations. Extensive experiments have been conducted to validate the superiority of DrRL on both recommendation accuracy and robustness.

In the realm of recommender systems (RS), Top-K ranking metrics such as NDCG@K are the gold standard for evaluating recommendation performance. However, during the training of recommendation models, optimizing NDCG@K poses significant challenges due to its inherent discontinuous nature and the intricate Top-K truncation. Recent efforts to optimize NDCG@K have either overlooked the Top-K truncation or suffered from high computational costs and training instability. To overcome these limitations, we propose SoftmaxLoss@K (SL@K), a novel recommendation loss tailored for NDCG@K optimization. Specifically, we integrate the quantile technique to handle Top-K truncation and derive a smooth upper bound for optimizing NDCG@K to address discontinuity. The resulting SL@K loss has several desirable properties, including theoretical guarantees, ease of implementation, computational efficiency, gradient stability, and noise robustness. Extensive experiments on four real-world datasets and three recommendation backbones demonstrate that SL@K outperforms existing losses with a notable average improvement of 6.03%. The code is available at https://github.com/Tiny-Snow/IR-Benchmark.

In order to integrate digital avatars into people’s lives, efficiently generating complete, realistic, and animatable avatars is crucial. However, increasing parameter counts and model sizes challenge the efficiency of training and deployment on devices. Additionally, traditional graphical rule-based micro-renderers, which simplify real-world photorealistic mechanisms such as illumination and reflections, fail to generate truly photorealistic images. To address these issues, we propose a two-stage model compression optimization architecture. In the first stage, our proposed distillation architecture compresses the model, and in the second stage, our generative adversarial renderer enhances the realism of digital avatars by customizing its inverse version to the student network. Specifically, during the knowledge distillation process, we achieve multi-scale feature fusion by concatenating the output features of RandLA-Net and GCN, combining global and local information to better capture the details and contextual information of the point cloud. We construct assisted supervision, enabling point-level supervision by building the graph topology of the entire point cloud. Furthermore, we propose feeding the extracted point cloud features as latent codes into our well-designed neural renderer to produce more realistic facial images. Experiments demonstrate that our method not only improves network performance but also significantly reduces the parameters and computation compared to existing state-of-the-art methods. Specifically, our method reduces the number of parameters of the teacher model by about 95% and the computation in knowledge distillation by 90%.

Digitally modeling and reconstructing talking humans is important in telepresence applications of AR or VR environments. However, current methods often fail to effectively address the inability to capture local details of avatars due to resolution or image quality limitations or effectively generate realistic and natural 3D color representations. Meanwhile, invisible regions may cause the reconstruction results to appear hollow or missing. To alleviate these problems, in this paper, we propose a novel approach, called GFAvatar. Compared to existing methods, GFAvatar improves the quality of point cloud texture features by designing the fusion of image texture and 3D texture information. We achieve end-to-end learning by using an image super-resolution approach combined with our designed PointNet variant to extract detailed features from head avatars and enhance the representation of point cloud features. Our multimodal color fusion network combines image and point cloud color data, generating more precise and expressive 3D color representations for better avatar quality. We also design a texture consistency loss function to address the problem of abnormal local color in the fusion network. Further, to efficiently address the challenges posed by disordered point clouds, we carefully elaborate a 3D grids optimization to improve the integrity of facial reconstruction. Extensive experimental results on available datasets indicate the superiority in comparison with most state-of-the-arts.

Uncertainty modeling and reasoning in intelligent systems are crucial for effective decision-making, such as complex evidence theory (CET) being particularly promising in dynamic information processing. Within CET, the complex basic belief assignment (CBBA) can model uncertainty accurately, while the complex rule of combination can effectively reason uncertainty with multiple sources of information, reaching a consensus. However, determining CBBA, as the key component of CET, remains an open issue. To mitigate this issue, we propose a novel method for generating CBBA using high-level features extracted from Box–Cox transformation and discrete Fourier transform (DFT). Specifically, our method deploys complex Gaussian fuzzy number (CGFN) to generate CBBA, which provides a more accurate representation of information. The proposed method is applied to pattern classification tasks through a multisource information fusion algorithm, and it is compared with several well-known methods to demonstrate its effectiveness. Experimental results indicate that our proposed CGFN-based method outperforms existing methods, by achieving the highest average classification rate in multisource information fusion for pattern classification tasks. We found the Box–Cox transformation contributes significantly to CGFN by formatting data in a normal distribution, and DFT can effectively extract high-level features. Our method offers a practical approach for generating CBBA in CET, precisely representing uncertainty and enhancing decision-making in uncertain scenarios.

The sparse interactions between users and items on the web have aggravated the difficulty of their representations in recommender systems. Existing approaches leverage item attributes (e.g., item categories and tags) to alleviate the data sparsity problem, so as to enhance the performance and interpretability of recommendation. However, directly using all attributes of items cannot avoid the negative impacts of low-quality attributes, where manually labeling the quality of attributes is time-consuming. To this end, we propose HQRec to jointly measure the quality of attributes automatically and perform recommendation accurately. Specifically, we first analyze the different qualities among item attributes, and propose to leverage item categories to select high-quality tags via category-guided quality measurement and direction-aware optimization in an unsupervised fashion. Then, we propose to capture the complex relations among users and items based on the high-quality attributes, where a novel quality-aware embedding fusion and quality-aware embedding propagation mechanism for users and items is devised. Extensive experiments on four real-world benchmark datasets show drastic performance gains brought by our proposed HQRec framework, which constantly achieves an average of 14.73% improvement over the state-of-the-art baselines in terms of Recall and NDCG metrics. Insightful case studies also show that our automatic quality measurements are highly accurate and interpretable.