The integration of large language models into open-world detection frameworks significantly improves versatility in new environments. Prompt representations derived from these models help establish classification boundaries for both base and novel ...

Despite superior performance on various natural language processing tasks, pre-trained models such as BERT are challenged by deploying on resource-constraint devices. Most existing model compression approaches require re-compression or fine-...

• YOCO-BERT achieves 2.1%–4.5% average accuracy improvement on the GLUE benchmark.
• The training complexity of YOCO-BERT is O ( 1 ) for N different devices.
• The architectures obtained by YOCO- BERT achieve better trade-off between ...

Rendering redundancy consumes a lot of computing resources in mobile devices. Eliminating redundancy can effectively improve system energy efficiency. However, as the premise of redundancy elimination, the existing redundancy identification methods ...

Label distribution learning (LDL) explicitly models label ambiguity based on the degree to which each label describes an instance. Existing LDL algorithms typically consider all features shared by all class labels. However, irrelevant ...

• By embedding a derived label-independent prediction space in the feature space and the label distribution space, we propose a Two-Stage label distribution ...

In large-scale recommender systems, retrieving top N relevant candidates accurately with resource constrain is crucial. To evaluate the performance of such retrieval models, Recall@N, the frequency of positive samples being retrieved in the top N ...

Recently, Non-parallel voice conversion (VC) has attracted the attention of many researchers in the field of speech. However, such model suffers from the limitation that how to improve the generalization of the model to extract the speaker ...

Multi-task learning (MTL) is a promising research field of machine learning, in which the training process of the neural network is equivalent to multi-objective optimization. On one hand, MTL trains all the network weights simultaneously to ...

Neural architecture search (NAS) has shown encouraging results in automating the architecture design. Recently, DARTS relaxes the search process with a differentiable formulation that leverages weight-sharing and SGD where all candidate operations are ...

By transferring both features and gradients between different layers, shortcut connections explored by ResNets allow us to effectively train very deep neural networks up to hundreds of layers. However, the additional computation costs induced by those ...

Learning feature interactions is crucial for click-through rate (CTR) prediction in recommender systems. In most existing deep learning models, feature interactions are either manually designed or simply enumerated. However, enumerating all feature ...

Online learning is used in a wide range of real applications, e.g., predicting ad click-through rates (CTR) and personalized recommendations. Based on the analysis of users' behaviors in Video-On-Demand (VoD) recommender systems,we discover that the ...

Line drawings are widely used in conveying shape and meaning for a broad range of objects with a minimum visual representation of a scene. As for large scale animated skinned meshes, efficiency and the quality of surface deformation matters the most. ...

A multimodel based range query processing algorithm is proposed to solve the information collection task for the CPSs, which utilizes multiple probability models to depict the data distribution of a sensor node. The execution of the multimodel based ...

Toy blocks can help the children develop various skills, such as spatial, mathematical, creative problem solving etc. In this paper, we developed a computer aided system for child to play blocks with a computer in a natural and intuitive way using the ...

Releasing sky lanterns is a popular way of celebrating festivals and ceremonies in the Asian countries. This paper presents a computer-aided approach to help novice users to design flyable sky lantern with desired shape. Given a closed up-right 3D model ...

One important research topic for the wireless sensor network is about the range query, i.e., retrieving the areas in the sensor network where the temperature is between 30 äï C and 35äï C. The traditional approaches are not feasible for such query since ...

To solve the multi-dimensional event based query in wireless sensor networks, this paper proposes four bloom filter based query processing algorithms UBP, BBP, SRBP and PBP. The four algorithms proposed can be classified into two classes: two bloom ...

Solving the eigenvalues of matrices is an open problem which is often related to scientific computation. With the increasing of the order of matrices, traditional sequential algorithms are unable to meet the needs for the calculation time. Although ...

In this paper a novel energy efficient algorithm based on relevant node selection is proposed to deal with a new kind of approximate query, named node number constraint query, for wireless sensor networks. The query processing algorithm is composed of ...

Event-based ad hoc query processing is one of the most important functions of wireless sensor networks. Data centric storage algorithm is carried out to solve the problem but the traditional algorithm neglects that events will last for a time. During ...