Search

From 2015, China began to promote eco-sustainability in the Belt and Road Initiative (BRI) through not only vision statements but also specific guidelines and governance initiatives. What has driven these policy changes? Bringing together theories of norm localization, norm subsidiarity, and policy deliberation, we argue that China’s move toward green BRI began as a norm localization process where environmental norms emerged in the open policy space created by China’s top leaders carrying the ambition to make the BRI a new global governance model. After adopting a broad norm on environmental stewardship, state bureaucracies found opportunities to create procedural and operational rules. A novel procedural rule-making methodology emphasizing inclusive dialogue with host countries has emerged, driven by top leaders’ pursuit of international leadership and preexisting local norms guiding South–South cooperation. With operational rules, different actors follow their preferences to localize existing international standards or develop new ones.In 2013, the Chinese government launched the Belt and Road Initiative (BRI) as a platform for global and regional cooperation focusing on infrastructure development, trade, finance, policy coordination, and people-to-people exchange. Since then, many researchers have raised concerns regarding negative environmental impacts of large-scale infrastructure development brought by the initiative and stressed the need for China to introduce stronger regulatory frameworks that promote ecological sustainability (hereinafter “eco-sustainability”) within the BRI (Ascensão et al. 2018; Lechner et al. 2018). Since 2015, China has made many efforts to govern the BRI’s environmental impacts, which are reflected at three levels, including setting broad policy goals, developing governance methodologies, and making operational rules and standards.First, the central government published several high-level documents, most notably the “Visions and Actions on Jointly Building Silk Road Economic Belt and 21st Century Maritime Silk Road” (2015) (hereinafter “Visions and Actions”), that show the goal to treat environmental stewardship as an indispensable target of the BRI (Coenen et al. 2021; Zhou et al. 2018). Second, regarding the methodology to design substantial rules and policies, China has stressed a “dialogue-based” approach, anchored on inclusive collective deliberation and innovation by China, host countries, and other international stakeholders (Tao 2020). A typical example of this approach is the BRI International Green Development Coalition (BRIGC), a policy dialogue platform cohosted by the Ministry of Ecology and Environment (MEE) and international partners from both the Global South and the Global North. By emphasizing the voices of and cooperation between developing countries, providing host country governments with opportunities to participate in the decision-making processes, and allowing flexibility in the rule adoption, this rule-making methodology differs from that of the West.1 Third, at the operational level, we are observing debates and uncertainties regarding some critical issues for environmental governance of the BRI, such as new areas of cooperation, standards for project design and implementation, and rules of development financing.Hence, Beijing’s green BRI policy seems not just a rhetorical strategy. Instead, by building new institutions and redesigning projects, China has made a clear shift in policy. China’s rapid move toward green BRI seems to contrast with the limited attention it had given to environmental impacts of its overseas activities in the past. Through the lens of constructivist international relations theories, Beijing’s move toward green BRI can be seen as a process of engagement with norms regarding appropriate ways to govern environmental impacts of international development cooperation and investment adopted by Western actors like the World Bank and the Organization for Economic Co-operation and Development (OECD) (Park 2005).2 In this process of normative engagement, Chinese policy makers have not only adopted key elements of some international norms—a “norm localization” dynamic—but also actively incorporated Chinese ideas, a “norm subsidiarity” dynamic (Acharya 2011).What has led China to gradually develop this green BRI policy, which concerns not only the import of existing environmental norms but also attempts to construct new ones? We argue that different types of norm dynamics exist in three phases of China’s green BRI policy. Initially, the green BRI development was driven by the localization of the international norm on environmental stewardship due to Chinese top leaders’ ambition to make the BRI a flagship program of international cooperation; after this broad goal had been set, the norm subsidiarity dynamic kicked in, which was largely driven by top leaders’ strategic ambition and preexisting local norms guiding South–South cooperation; and finally, for the development of operational rules and standards, different bureaucracies have followed their preferences to localize existing international norms or develop new norms.3We proceed as follows. After reviewing the literature on green BRI, we introduce our analytical framework building on Acharya’s (2011) framework of norm localization and norm subsidiarity. Considering China’s political context, we further stress the importance of local deliberation—a process of collective policy deliberation among top leaders and political actors in functional areas—in triggering the localization dynamic and (potentially) transforming the localization dynamic into the subsidiarity dynamic. We then conduct a theory-guided process-tracing study on the evolution of China’s efforts to green the BRI. By unpacking the logic of China’s green BRI policy, we advance understanding of China’s strategy for governing environmental impacts of its global activities and also China’s engagement with international environmental norms.

Large language models (LLMs) often contain misleading content, emphasizing the need to align them with human values to ensure secur AI systems. Reinforcement learning from human feedback (RLHF) has been employed to achieve this alignment by combining a reward model, typically based on Bradley-Terry paired comparison, with an RL algorithm such as Proximal Policy Optimization (PPO) to optimize LLM responses. However, RLHF exhibits complexity, instability, and sensitivity to hyperparameters. In this paper, we propose Preference Ranking Optimization (PRO) as an alternative to PPO for directly aligning LLMs with the Bradley-Terry comparison. PRO extends the pairwise Bradley-Terry comparison to accommodate preference rankings of any length. By iteratively contrasting the likelihood of generating responses, PRO instructs the LLM to prioritize the best response while progressively ranking the remaining responses. In this manner, PRO effectively transforms human alignment into aligning the probability ranking of $n$ responses generated by LLM with the preference ranking of humans towards these responses. Experiments have shown that PRO outperforms existing alignment algorithms, achieving comparable results to ChatGPT and human responses through automatic-based, reward-based, GPT-4, and human evaluations. Furthermore, we demonstrate that longer, more diverse, and higher-quality preference ranking sequences can consistently enhance the performance of human alignment.

Large language models (LLMs) demonstrate remarkable ability to comprehend, reason, and generate following nature language instructions. However, the development of LLMs has been primarily focused on high-resource languages, such as English, thereby limiting their applicability and research in other languages. Consequently, we present PolyLM, a multilingual LLM trained on 640 billion (B) tokens, avaliable in two model sizes: 1.7B and 13B. To enhance its multilingual capabilities, we 1) integrate bilingual data into training data; and 2) adopt a curriculum learning strategy that increases the proportion of non-English data from 30% in the first stage to 60% in the final stage during pre-training. Further, we propose a multilingual self-instruct method which automatically generates 132.7K diverse multilingual instructions for model fine-tuning. To assess the model's performance, we collect several existing multilingual tasks, including multilingual understanding, question answering, generation, and translation. Extensive experiments show that PolyLM surpasses other open-source models such as LLaMA and BLOOM on multilingual tasks while maintaining comparable performance in English. Our models, alone with the instruction data and multilingual benchmark, are available at: \url{https://modelscope.cn/models/damo/nlp_polylm_13b_text_generation}.

Supplementary document.

Recently developed graph contrastive learning (GCL) approaches compare two different "views" of the same graph in order to learn node/graph representations. The core assumption of these approaches is that by graph augmentation, it is possible to generate several structurally different but semantically similar graph structures, and therefore, the identity labels of the original and augmented graph/nodes should be identical. However, in this paper, we observe that this assumption does not always hold, for example, any perturbation to nodes or edges in a molecular graph will change the graph labels to some degree. Therefore, we believe that augmenting the graph structure should be accompanied by an adaptation of the labels used for the contrastive loss. Based on this idea, we propose ID-MixGCL, which allows for simultaneous modulation of both the input graph and the corresponding identity labels, with a controllable degree of change, leading to the capture of fine-grained representations from unlabeled graphs. Experimental results demonstrate that ID-MixGCL improves performance on graph classification and node classification tasks, as demonstrated by significant improvements on the Cora, IMDB-B, and IMDB-M datasets compared to state-of-the-art techniques, by 3-29% absolute points., Comment: 9 pages, 6 figures

When trying to answer complex questions, people often rely on multiple sources of information, such as visual, textual, and tabular data. Previous approaches to this problem have focused on designing input features or model structure in the multi-modal space, which is inflexible for cross-modal reasoning or data-efficient training. In this paper, we call for an alternative paradigm, which transforms the images and tables into unified language representations, so that we can simplify the task into a simpler textual QA problem that can be solved using three steps: retrieval, ranking, and generation, all within a language space. This idea takes advantage of the power of pre-trained language models and is implemented in a framework called Solar. Our experimental results show that Solar outperforms all existing methods by 10.6-32.3 pts on two datasets, MultimodalQA and MMCoQA, across ten different metrics. Additionally, Solar achieves the best performance on the WebQA leaderboard, Comment: Findings of ACL 2023

Document-level Event Extraction (DEE) is particularly tricky due to the two challenges it poses: scattering-arguments and multi-events. The first challenge means that arguments of one event record could reside in different sentences in the document, while the second one reflects one document may simultaneously contain multiple such event records. Motivated by humans' reading cognitive to extract information of interests, in this paper, we propose a method called HRE (Human Reading inspired Extractor for Document Events), where DEE is decomposed into these two iterative stages, rough reading and elaborate reading. Specifically, the first stage browses the document to detect the occurrence of events, and the second stage serves to extract specific event arguments. For each concrete event role, elaborate reading hierarchically works from sentences to characters to locate arguments across sentences, thus the scattering-arguments problem is tackled. Meanwhile, rough reading is explored in a multi-round manner to discover undetected events, thus the multi-events problem is handled. Experiment results show the superiority of HRE over prior competitive methods., Comment: To apper in ICASSP2022

Open Information Extraction (OpenIE) facilitates domain-independent discovery of relational facts from large corpora. The technique well suits many open-world natural language understanding scenarios, such as automatic knowledge base construction, open-domain question answering, and explicit reasoning. Thanks to the rapid development in deep learning technologies, numerous neural OpenIE architectures have been proposed and achieve considerable performance improvement. In this survey, we provide an extensive overview of the-state-of-the-art neural OpenIE models, their key design decisions, strengths and weakness. Then, we discuss limitations of current solutions and the open issues in OpenIE problem itself. Finally we list recent trends that could help expand its scope and applicability, setting up promising directions for future research in OpenIE. To our best knowledge, this paper is the first review on this specific topic., Comment: Accepted by IJCAI22 survey track

In recent years, with the rapid growth of Internet data, the number and types of scientific and technological resources are also rapidly expanding. However, the increase in the number and category of information data will also increase the cost of information acquisition. For technology-based enterprises or users, in addition to general papers, patents, etc., policies related to technology or the development of their industries should also belong to a type of scientific and technological resources. The cost and difficulty of acquiring users. Extracting valuable science and technology policy resources from a huge amount of data with mixed contents and providing accurate and fast retrieval will help to break down information barriers and reduce the cost of information acquisition, which has profound social significance and social utility. This article focuses on the difficulties and problems in the field of science and technology policy, and introduces related technologies and developments., Comment: There are some problems in the conclusions and analysis of this paper, and we need to withdraw this paper

Supplemental material

With the rapid development of Internet technology, people have more and more access to a variety of web page resources. At the same time, the current rapid development of deep learning technology is often inseparable from the huge amount of Web data resources. On the other hand, NLP is also an important part of data processing technology, such as web page data extraction. At present, the extraction technology of web page text mainly uses a single heuristic function or strategy, and most of them need to determine the threshold manually. With the rapid growth of the number and types of web resources, there are still problems to be solved when using a single strategy to extract the text information of different pages. This paper proposes a web page text extraction algorithm based on multi-feature fusion. According to the text information characteristics of web resources, DOM nodes are used as the extraction unit to design multiple statistical features, and high-order features are designed according to heuristic strategies. This method establishes a small neural network, takes multiple features of DOM nodes as input, predicts whether the nodes contain text information, makes full use of different statistical information and extraction strategies, and adapts to more types of pages. Experimental results show that this method has a good ability of web page text extraction and avoids the problem of manually determining the threshold., Comment: Part of the content is for other purposes

Solving differential equations is a critical task in scientific computing. Domain-specific languages (DSLs) have been a promising direction in achieving performance and productivity, but the current state of the art only supports stencil computation, leaving solvers requiring loop-carried dependencies aside. Alternatively, sparse matrices can represent such equation solvers and are more general than existing DSLs, but the performance is sacrificed. This paper points out that sparse matrices can be represented as programs instead of data, having both the generality from the matrix-based representation and the performance from program optimizations. Based on the idea, we propose the Staged Sparse Row (SSR) sparse matrix representation that can efficiently cover applications on structured grids. With SSR representation, users can intuitively define SSR matrices using generator functions and use SSR matrices through a concise object-oriented interface. SSR matrices can then be chained and applied to construct the algorithm, including those with loop-carried dependences. We then apply a set of dedicated optimizations, and ultimately simplify the SSR matrix-based codes into straightforward matrix-free ones, which are efficient and friendly for further analysis. Implementing BT pseudo application in the NAS Parallel Benchmark, with less than $10\%$ lines of code compared with the matrix-free reference FORTRAN implementation, we achieved up to $92.8\%$ performance. Implementing a matrix-free variant for the High-Performance Conjugate Gradient benchmark, we achieve $3.29\times$ performance compared with the reference implementation, while our implementation shares the same algorithm on the same programming abstraction, which is sparse matrices.

Traditional photonic integrated circuit (PIC) inherits the mature CMOS fabrication process from the electronic integrated circuit (IC) industry. However, this process also limits the PIC structure to a single-waveguide-layer configuration. In this work, we explore the possibility of the multi-waveguide-layer PIC by proposing and demonstrating a true 3-D optical phased array (OPA) device, with the light exiting from the edge of the device, based on a multi-layer Si3N4/SiO2 platform. The multi-waveguide-layer configuration offers the possibility of utilizing edge couplers at both the input and the emitting ends to achieve broadband high efficiency. This uniqueness provides the potential for a more extended detection range in the Lidar application. The device has been studied by numerical simulation, and proof-of-concept samples have been fabricated and tested with a CMOS-compatible process. To the best of our knowledge, this is the first experimental proof-of-concept of a true 3-D OPA with a multi-waveguide-layer configuration all over the device.

Lifelong learning aims to accumulate knowledge and alleviate catastrophic forgetting when learning tasks sequentially. However, existing lifelong language learning methods only focus on the supervised learning setting. Unlabeled data, which can be easily accessed in real-world scenarios, are underexplored. In this paper, we explore a novel setting, semi-supervised lifelong language learning (SSLL), where a model learns sequentially arriving language tasks with both labeled and unlabeled data. We propose an unlabeled data enhanced lifelong learner to explore SSLL. Specially, we dedicate task-specific modules to alleviate catastrophic forgetting and design two modules to exploit unlabeled data: (1) a virtual supervision enhanced task solver is constructed on a teacher-student framework to mine the underlying knowledge from unlabeled data; and (2) a backward augmented learner is built to encourage knowledge transfer from newly arrived unlabeled data to previous tasks. Experimental results on various language tasks demonstrate our model's effectiveness and superiority over competitive baselines under the new setting SSLL., Comment: EMNLP Findings 2022 Long Paper

Thermal radiation originates from photons which are spontaneously emitted by different points in a thermal source. All matter with a temperature greater than 0K emits thermal radiation, which in turn can extract temperature distribution for all optical and electronic materials and devices. The classic model was found by Max Planck, which describes the spectral density of electromagnetic radiation emitted by a blackbody in the thermal equilibrium at a given temperature. A blackbody is an ideal emitter which absorbs all incident electromagnetic radiation from all frequencies and angles of incidence. All other emissions are studied with the reference of the blackbody following the properties of emission and absorption. Emissivity is core in thermal devices, both source and receiver. Thermal emission is a fundamental phenomenon for which any object with temperature greater than absolute zero emits energy in the form of electromagnetic waves. The thermal emission from a blackbody (an ideal thermal emitter) only depends on its temperature, as described by Planck’s law. For a nonideal thermal emitter, i.e., a gray body, thermal emission also depends on how effectively the emitter can radiate, quantified as its emissivity.In this dissertation, I systematically investigate the thermal emission from nanoplamonic structures in both near- and far-field regimes such as metasurfaces, 2D materials, single plasmonic emitter, etc. I implement numerical simulations to investigate the directional control of narrow-band perfect thermal emission using a nanoscale Yagi–Uda antenna consisting of an array of nanowires. Then I fabricate single plasmonic device and plasmonic structures via the state-of-the-art micro and nanofabrication techniques including electron beam (e-beam) lithography, photolithography, e-beam evaporation, reactive ion etching (RIE), wet etching, focused ion beam (FIB), etc. Finally, I demonstrate the unique properties of the nanoplasmonic structures we fabricate via thermal and optical characterization.

Few-shot classification tends to struggle when it needs to adapt to diverse domains. Due to the non-overlapping label space between domains, the performance of conventional domain adaptation is limited. Previous work tackles the problem in a transductive manner, by assuming access to the full set of test data, which is too restrictive for many real-world applications. In this paper, we set out to tackle this issue by introducing a inductive framework, DaFeC, to improve Domain adaptation performance for Few-shot classification via Clustering. We first build a representation extractor to derive features for unlabeled data from the target domain (no test data is necessary) and then group them with a cluster miner. The generated pseudo-labeled data and the labeled source-domain data are used as supervision to update the parameters of the few-shot classifier. In order to derive high-quality pseudo labels, we propose a Clustering Promotion Mechanism, to learn better features for the target domain via Similarity Entropy Minimization and Adversarial Distribution Alignment, which are combined with a Cosine Annealing Strategy. Experiments are performed on the FewRel 2.0 dataset. Our approach outperforms previous work with absolute gains (in classification accuracy) of 4.95%, 9.55%, 3.99% and 11.62%, respectively, under four few-shot settings., Comment: Accepted by ECML-PKDD 2020

According to the application of fiber channel in electronic systems for aerospace, FC-AE-1553 protocol draws 1553B bus command/response type of communication, and also has an excellent network performance of fiber channel. This paper gives the details of protocol conversion from MIL-STD-1553B to FC-AE-1553, and system design of the network controller NC and the network terminal NT. In addition, the performance analysis for the system is expressed. This system is not only compatible with the original 1553B terminal equipment, but also can be well integrated with fiber channel technology, and it has a far-reaching significance on overall upgrading of aerospace electronic systems.