Search

Text-based delivery addresses, as the data foundation for logistics systems, contain abundant and crucial location information. How to effectively encode the delivery address is a core task to boost the performance of downstream tasks in the logistics system. Pre-trained Models (PTMs) designed for Natural Language Process (NLP) have emerged as the dominant tools for encoding semantic information in text. Though promising, those NLP-based PTMs fall short of encoding geographic knowledge in the delivery address, which considerably trims down the performance of delivery-related tasks in logistic systems such as Cainiao. To tackle the above problem, we propose a domain-specific pre-trained model, named G2PTL, a Geography-Graph Pre-trained model for delivery address in Logistics field. G2PTL combines the semantic learning capabilities of text pre-training with the geographical-relationship encoding abilities of graph modeling. Specifically, we first utilize real-world logistics delivery data to construct a large-scale heterogeneous graph of delivery addresses, which contains abundant geographic knowledge and delivery information. Then, G2PTL is pre-trained with subgraphs sampled from the heterogeneous graph. Comprehensive experiments are conducted to demonstrate the effectiveness of G2PTL through four downstream tasks in logistics systems on real-world datasets. G2PTL has been deployed in production in Cainiao's logistics system, which significantly improves the performance of delivery-related tasks.

Large-scale commercial platforms usually involve numerous business domains for diverse business strategies and expect their recommendation systems to provide click-through rate (CTR) predictions for multiple domains simultaneously. Existing promising and widely-used multi-domain models discover domain relationships by explicitly constructing domain-specific networks, but the computation and memory boost significantly with the increase of domains. To reduce computational complexity, manually grouping domains with particular business strategies is common in industrial applications. However, this pre-defined data partitioning way heavily relies on prior knowledge, and it may neglect the underlying data distribution of each domain, hence limiting the model's representation capability. Regarding the above issues, we propose an elegant and flexible multi-distribution modeling paradigm, named Adaptive Distribution Hierarchical Model (AdaptDHM), which is an end-to-end optimization hierarchical structure consisting of a clustering process and classification process. Specifically, we design a distribution adaptation module with a customized dynamic routing mechanism. Instead of introducing prior knowledge for pre-defined data allocation, this routing algorithm adaptively provides a distribution coefficient for each sample to determine which cluster it belongs to. Each cluster corresponds to a particular distribution so that the model can sufficiently capture the commonalities and distinctions between these distinct clusters. Extensive experiments on both public and large-scale Alibaba industrial datasets verify the effectiveness and efficiency of AdaptDHM: Our model achieves impressive prediction accuracy and its time cost during the training stage is more than 50% less than that of other models.

We report a high-beam-quality long pulse 532 nm laser, which can operate in wide range of temperature. By using a master-oscillator power amplifier (MOPA) system and LBO extracavity-frequency-doubling, we achieved 532 nm laser with the maximum output power of 32 W, the pulse frequency of 12 kHz, and the pulse width 90.2 ns. In addition, the beam quality factors reached M2 x = 1 . 31 , M2 y = 1 . 36 , respectively. In order to get long pulse, apart from choosing suitable transmissivity of output coupler, we also spreaded the cavity length with a 4f imaging system. The laser was proved to be stable when the ambient temperature changed from −19 °C to 27 °C. Compared to other green lasers, this laser can be applied in Rayleigh beaconing because of its long pulse and wide range of working temperature.

We report on the development and application of a novel, to the best of our knowledge, all-solid-state tunable narrow-linewidth 226 nm UV laser system. The laser system consists of three parts: a tunable single-frequency Ti:sapphire 787 nm laser, a single-frequency long-pulse-width flattop-shaped 532 nm laser, and a nonlinear frequency transformation system. The 532 nm laser is a sum-frequency mixed with the second harmonic of the 787 nm laser to produce the 226 nm laser. The maximum output pulse energy at 226 nm is 3 mJ. Nitric oxide planar laser-induced fluorescence velocimetry is demonstrated in the China Aerodynamics Research and Development Center’s FD14 hypersonic shock tunnel using this 226 nm laser system. It is proven that this laser is convenient for high-resolution molecular tagging fluorescence spectroscopy.

A 2-μm composite Tm:YAG laser pumped with a narrow-band laser diode was presented. The temperature distribution and thermal lens in the Tm:YAG were numerically simulated by a finite element method and the results were used to design the special cavity, in order to achieve a high efficiency and stable output. With a 25-W incident pump power, we obtained a maximum output power of 11 W at 2018.5 nm, corresponding to a slope efficiency of 51.3% and an optical-to-optical efficiency of 44.5%, respectively. The beam quality was measured to be and .