Contrastive Transformer Hashing for Compact Video Representation.

Video hashing learns compact representation by mapping video into low-dimensional Hamming space and has achieved promising performance in large-scale video retrieval. It is challenging to effectively exploit temporal and spatial structure in an unsupervised setting. To fulfill this gap, this paper proposes Contrastive Transformer Hashing (CTH) for effective video retrieval. Specifically, CTH develops a bidirectional transformer autoencoder, based on which visual reconstruction loss is proposed. CTH is more powerful to capture bidirectional correlations among frames than conventional unidirectional models. In addition, CTH devises multi-modality contrastive loss to reveal intrinsic structure among videos. CTH constructs inter-modality and intra-modality triplet sets and proposes multi-modality contrastive loss to exploit inter-modality and intra-modality similarities simultaneously. We perform video retrieval tasks on four benchmark datasets, i.e., UCF101, HMDB51, SVW30, FCVID using the learned compact hash representation, and extensive empirical results demonstrate the proposed CTH outperforms several state-of-the-art video hashing methods.

Physics-informed neural networks to solve lumped kinetic model for chromatography process.

Numerical method is widely used for solving the mechanistic models of chromatography process, but it is time-consuming and hard to response in real-time. Physics-informed neural network (PINN) as an emerging technology combines the structure of neural network with physics laws, and is getting noticed for solving physics problems with a balanced accuracy and calculation speed. In this research, a proof-of-concept study was carried out to apply PINN to chromatography process simulation. The PINN model structure was designed for the lumped kinetic model (LKM) with all LKM parameters. The PINN structure, training data and model complexity were optimized, and an optimal mode was obtained by adopting an in-series structure with a nonuniform training data set focusing on the breakthrough transition region. A PINN for LKM (LKM-PINN) consisting of four neural networks, 12 layers and 606 neurons was then used for the simulation of breakthrough curves of chromatography processes. The LKM parameters were estimated with two breakthrough curves and used to infer the breakthrough curves at different residence times, loading concentrations and column sizes. The results were comparable to that obtained with numerical methods. With the same raw data and constraints, the average fitting error for LKM-PINN model was 0.075, which was 0.081 for numerical method. With the same initial guess, the LKM-PINN model took 160 s to complete the fitting, while the numerical method took 7 to 72 min, depending on the fitting settings. The fitting speed of LKM-PINN model was further improved to 30 s with random initial guess. Thus, the LKM-PINN model developed in this study is capable to be applied to real-time simulation for digital twin.

Semi-Paired Discrete Hashing: Learning Latent Hash Codes for Semi-Paired Cross-View Retrieval.

Due to the significant reduction in computational cost and storage, hashing techniques have gained increasing interests in facilitating large-scale cross-view retrieval tasks. Most cross-view hashing methods are developed by assuming that data from different views are well paired, e.g., text-image pairs. In real-world applications, however, this fully-paired multiview setting may not be practical. The more practical yet challenging semi-paired cross-view retrieval problem, where pairwise correspondences are only partially provided, has less been studied. In this paper, we propose an unsupervised hashing method for semi-paired cross-view retrieval, dubbed semi-paired discrete hashing (SPDH). In specific, SPDH explores the underlying structure of the constructed common latent subspace, where both paired and unpaired samples are well aligned. To effectively preserve the similarities of semi-paired data in the latent subspace, we construct the cross-view similarity graph with the help of anchor data pairs. SPDH jointly learns the latent features and hash codes with a factorization-based coding scheme. For the formulated objective function, we devise an efficient alternating optimization algorithm, where the key binary code learning problem is solved in a bit-by-bit manner with each bit generated with a closed-form solution. The proposed method is extensively evaluated on four benchmark datasets with both fully-paired and semi-paired settings and the results demonstrate the superiority of SPDH over several other state-of-the-art methods in term of both accuracy and scalability.