人民财讯1月8日电，广州市人民政府办公厅印发《广州市加快建设先进制造业强市规划(2024—2035 年)》。其中提出，加快研发具有自主知识产权的操作系统、数据库、中间件、办公软件等通用基础软 件，提高产品兼容性，加快构建新型自主产业生态。聚焦CAD/CAM/CAE等工业通用工具软件、EDA技 术研发与应用、工业软件共性支撑技术、重大工程与特色行业软件等方面，推动建立新的工业软件标 准，掌握自主知识产权，取得若干标志性成果，统筹推进自主创新工业操作系统、中间件、工业APP、 新型数据库管理系统的全面应用。开展工业APP开发与应用创新，推动优秀工业APP及应用解决方案在 行业内的推广应用。布局下一代云计算软件体系，提升云安全水平和智能云服务能力，支持高性能采 集、高容量存储、海量信息处理、人工智能算法、工具群、区块链等技术创新。加快发展新型机器学 习、虚拟现实、元宇宙相关新兴平台软件。 ...

Group 1 - Michael I. Jordan is recognized as a pioneer in machine learning and has recently been elected as a foreign academician of the Chinese Academy of Sciences [1][17] - He has extensive experience in both academia and industry, having collaborated with companies like Ant Group and Amazon, and has been involved in founding multiple companies [1][24] - Jordan emphasizes the importance of curiosity and applying mathematical thinking to solve real-world problems [1][2][27] Group 2 - Jordan's work spans various fields, including decision-making, knowledge exchange, and data prediction in real-world scenarios, which he believes does not require extraordinary talent but rather dedication and hard work [3][18] - He highlights the close relationship between machine learning and statistics, stating that learning involves making predictions based on statistical methods [7][22] - Jordan has engaged in significant collaborations in China, particularly in meteorology, where he worked on predicting severe weather events using machine learning [8][23] Group 3 - He has been involved in the development of tools that are currently used by Chinese meteorological departments, showcasing the practical applications of his research [8][23] - Jordan believes there is no disconnect between industry development and academic research, as many advancements in technology stem from academic findings [10][25] - He notes that while China has made strides in open-source initiatives, there is still a need to focus on creativity and problem-solving skills rather than just academic metrics [11][27]

Core Insights - The South African Revenue Service (SARS) plans to intensify its crackdown on tax non-compliance by 2026, with a focus on scrutinizing taxpayers' bank accounts as a primary enforcement tool [1] Tax Revenue Performance - In the 2024/25 fiscal year, SARS collected a record 2.303 trillion rand in taxes, with refunds amounting to 447.3 billion rand, reflecting an 8.1% year-on-year increase [1] - The growth in net personal income tax revenue is attributed to increased withholding taxes from sectors such as financial services, real estate, and business services, as well as higher-than-expected withdrawals under the two-bucket pension system [1] Compliance Enforcement - Despite improved fiscal revenue, SARS emphasizes that compliance enforcement remains a top priority, generating 304 billion rand from compliance initiatives in the 2024/25 fiscal year, a nearly 17% year-on-year increase [1] - Of this, 156.1 billion rand was collected through direct recovery, while 147.9 billion rand was aimed at preventing tax base erosion [1] Technological Integration - SARS is leveraging artificial intelligence, data science, and machine learning to effectively identify non-compliance by analyzing taxpayer transactions and banking data [1] - Under the Tax Administration Act, SARS has the authority to access bank and cryptocurrency information and can directly deduct amounts from accounts in cases of tax arrears [1] Future Outlook - Experts predict that bank account scrutiny will continue to be a significant tool for tax enforcement in 2026 [1]

Core Insights - QYResearch is recognized as a strategic partner rather than just a data provider, emphasizing its role in decision-making support for clients across various industries [3][7][8] - The company has established long-term relationships with clients, providing tailored solutions that enhance strategic planning and market insights [4][6][7] Group 1: Client Relationships and Feedback - QYResearch maintains long-term partnerships with leading companies, offering detailed market data and strategic decision support through customized solutions [3][4] - Clients, such as a multinational medical device company, highlight the value of QYResearch's insights in avoiding risks and seizing opportunities [3][4] - A cross-national consumer electronics company praised QYResearch for providing not just data but actionable insights and solutions in a complex market [4][6] Group 2: Service Capabilities - The company offers a comprehensive range of services, from standardized market research reports to customized analyses and competitor assessments [3][5] - QYResearch's ability to integrate and analyze data across regions and industries allows it to provide in-depth insights into global markets [3][5] - The firm employs advanced technologies, including AI and big data analytics, to enhance the efficiency and accuracy of its market insights [4][5] Group 3: Customization and Flexibility - QYResearch tailors its research solutions to meet the diverse strategic needs of clients, whether for short-term projects or long-term collaborations [5][6] - An example includes a leading consumer electronics company in South Korea that received customized market analysis and strategy development support [6] Group 4: Building Trust and Long-term Value - The company focuses on establishing trust with clients, ensuring that research outcomes align with their strategic goals through continuous communication [7] - Long-term clients appreciate QYResearch's proactive approach in providing forward-looking recommendations that enhance strategic execution [7][8] - QYResearch's commitment to professional, precise, and customized services has earned it high praise and long-term trust from clients [7][8]

Quantitative Models and Construction Methods Model Name: mHC-MLP - **Model Construction Idea**: The mHC-MLP model introduces manifold-constrained hyper-connections (mHC) into the traditional MLP framework to address issues such as low signal-to-noise ratio, non-stationarity, and extreme tail behavior in financial data. It achieves this by incorporating multi-stream residual channels, gated fan-in/fan-out mappings, and doubly stochastic manifold projections (via Sinkhorn-Knopp) to enhance numerical stability and extrapolation resistance[1][16][22]. - **Model Construction Process**: 1. **Multi-Stream Residual Channels**: The model expands the single residual channel in traditional ResNet to multiple parallel sub-streams, allowing independent feature representations and dynamic routing between streams[19][20]. 2. **Manifold Constraints**: - Residual mixing matrices are constrained to the Birkhoff polytope (doubly stochastic matrices), ensuring non-negativity, row sums of 1, and column sums of 1. This is achieved using the Sinkhorn-Knopp algorithm during training[22][23][54]. - Fan-in and fan-out mappings are constrained to non-negative values using sigmoid functions, ensuring that output features remain within the convex hull of input features[24]. 3. **Dynamic Routing Mechanism**: The model uses a combination of linear mixing (via residual matrices) and non-linear transformations (via MLP blocks) to balance feature interaction and noise suppression[49][50][51]. 4. **Deep Stacking**: The mHC-MLP extends the network depth to six layers, leveraging the numerical stability provided by manifold constraints to capture higher-order interactions[56][57]. 5. **Initialization and Regularization**: Parameters are initialized with minimal values (e.g., alpha = 0.01) to ensure stable gradient flow during early training stages. Regularization is achieved through manifold constraints rather than traditional dropout or L2 regularization[25][55]. - **Model Evaluation**: The mHC-MLP model demonstrates improved numerical stability, reduced overfitting, and enhanced robustness against noise. However, it may underperform in short-term, high-volatility scenarios due to its conservative nature[2][75][86]. --- Model Backtesting Results mHC-MLP Model - **Cumulative Return**: 49% (compared to 56% for the unconstrained MLP model)[75] - **t-Statistic**: Not explicitly mentioned for mHC-MLP - **IC_IR**: Not explicitly mentioned for mHC-MLP - **Turnover**: Lower than the unconstrained MLP model, indicating better stability[2][75] - **Maximum Drawdown**: Lower than the unconstrained MLP model, reflecting reduced risk exposure[2][75] --- Quantitative Factors and Construction Methods Factor Name: Barra MLP Factor - **Factor Construction Idea**: The Barra MLP factor leverages neural networks to capture non-linear interactions and complex relationships between Barra style factors and residual stock returns, overcoming the limitations of traditional linear factor models[30][31]. - **Factor Construction Process**: 1. **Baseline Risk Model**: A long-term risk model is constructed using the Barra CNE6 framework, incorporating one country factor, 31 industry factors, and 15 style factors (e.g., size, beta, momentum, value)[36][37][38]. 2. **Residual Return Extraction**: Stock returns are decomposed into common factor contributions and residual returns via cross-sectional regression. The residual returns serve as the prediction target for the MLP model[40]. 3. **Rolling Training**: The MLP model is trained using rolling windows of 24, 36, and 72 months to balance bias and variance. Features include the 15 style factors, and the target is the next-period residual return[41]. 4. **Multi-Period Signal Synthesis**: Predictions from the three training windows are standardized (Z-score) and combined using equal weighting or IC-based weighting to generate a composite factor[42][43]. 5. **Orthogonalization**: The composite factor is regressed against the 15 style factors to remove linear correlations, ensuring it provides incremental information[44]. 6. **Pure Factor Return Calculation**: The orthogonalized factor is incorporated into an enhanced Barra risk model, and its pure factor return is estimated via cross-sectional regression[45]. - **Factor Evaluation**: The Barra MLP factor effectively captures non-linear alpha signals and demonstrates significant cumulative returns and IC_IR values, validating its utility in quantitative strategies[46]. --- Factor Backtesting Results Barra MLP Factor - **Cumulative Return**: Over 15%[46] - **t-Statistic**: 2.8[46] - **IC_IR**: 0.45[46] - **Turnover**: Not explicitly mentioned - **Maximum Drawdown**: Not explicitly mentioned --- Composite Model: mHC-Enhanced Barra MLP Factor - **Model Construction Idea**: The mHC-enhanced Barra MLP factor integrates the mHC architecture into the Barra MLP framework to improve robustness and stability while retaining the ability to capture non-linear interactions[48]. - **Model Construction Process**: The MLP core in the Barra MLP factor is replaced with the mHC-MLP architecture, maintaining the same input features, target variables, and training framework. This modification introduces manifold constraints and dynamic routing to enhance numerical stability and reduce overfitting[48][49][50]. - **Model Evaluation**: While the mHC-enhanced factor demonstrates superior stability and robustness, it may lag in short-term, high-volatility markets due to its conservative design[75][86]. --- Composite Model Backtesting Results mHC-Enhanced Barra MLP Factor - **Cumulative Return**: Not explicitly mentioned - **t-Statistic**: Not explicitly mentioned - **IC_IR**: Not explicitly mentioned - **Turnover**: Lower than the original Barra MLP factor[2][75] - **Maximum Drawdown**: Lower than the original Barra MLP factor[2][75]

本文首发于公众号"私募排排网"。 （点击↑↑ 上图查看详情 ） 近年来 ， 私募行业 一股新的投资力量正在迅速崛起 ， 成为业内焦点 ——量化宏观策略 。 以往宏观策略多为主观多头出身的管理人，而今年 量化背景机构的宏观策略产品一度爆火。 在全球知名对冲基金中，宏观策略已成为各自的核心策略，如桥水、城堡、AQR、元盛等均已深度布局，也有越来越多的新兴对冲基金开始将 量化方法应用于宏观投资。根据Preqin数据，2020年以来采用量化宏观策略的对冲基金管理规模年均增长超过15%，远超传统主观宏观策略的增 长速度。 业绩表现来看，截至11月底，有业绩显示的195只宏观策略产品，今年来收益均值为25.50%，其中主观宏观策略产品收益均值为26.42%、量化 宏观策略产品收益均值为21.42%。虽在业绩表现上，量化宏观略有逊色，但从持有体验上来看，量化宏观优势凸显：量化宏观策略产品今年来 的夏普均值高达2.11，而主观宏观策略产品的夏普均值为1.57。 01 为什么量化宏观策略突然火爆？ 而部分量化宏观策略却凭借对市场流动性的实时监控和压力测试模型成功规避了最严重的损失，这一鲜明对比引发了行业对投资方法论的深思。 比 ...

Quantitative Models and Construction Methods 1. Model Name: GAN_GRU - **Model Construction Idea**: The GAN_GRU model combines Generative Adversarial Networks (GAN) for feature generation and Gated Recurrent Unit (GRU) for time-series feature encoding to construct a stock selection factor[4][13][14] - **Model Construction Process**: 1. **GAN Component**: - The generator (G) learns the real data distribution and generates realistic samples from random noise \( z \) (Gaussian or uniform distribution). The generator's loss function is: $$ L_{G} = -\mathbb{E}_{z\sim P_{z}(z)}[\log(D(G(z)))] $$ where \( D(G(z)) \) represents the discriminator's probability of classifying generated data as real[24][25][26] - The discriminator (D) distinguishes real data from generated data. Its loss function is: $$ L_{D} = -\mathbb{E}_{x\sim P_{data}(x)}[\log D(x)] - \mathbb{E}_{z\sim P_{z}(z)}[\log(1-D(G(z)))] $$ where \( D(x) \) is the probability of real data being classified as real, and \( D(G(z)) \) is the probability of generated data being classified as real[27][29][30] - GAN training alternates between optimizing \( G \) and \( D \) until convergence[30] 2. **GRU Component**: - Two GRU layers (GRU(128, 128)) are used to encode time-series features, followed by a Multi-Layer Perceptron (MLP) with layers (256, 64, 64) to predict returns. The final output \( pRet \) is used as the stock selection factor[22] 3. **Feature Input and Processing**: - Input features include 18 price-volume characteristics (e.g., closing price, turnover, etc.) sampled over the past 400 days, with a shape of \( 40 \times 18 \) (40 days of features)[18][19][37] - Features undergo outlier removal, standardization, and cross-sectional normalization[18] 4. **Training Details**: - Training-validation split: 80%-20% - Semi-annual rolling training (June 30 and December 31 each year) - Hyperparameters: batch size equals the number of stocks, Adam optimizer, learning rate \( 1e-4 \), IC loss function, early stopping (10 rounds), max training rounds (50)[18] 5. **Stock Selection**: - Stocks are filtered to exclude ST stocks and those listed for less than six months[18] - **Model Evaluation**: The GAN_GRU model effectively captures price-volume time-series features and demonstrates strong predictive power for stock returns[4][13][22] --- Model Backtesting Results 1. GAN_GRU Model - **IC Mean**: 0.1119*** (2019-2025)[4][41] - **ICIR (non-annualized)**: 0.89[42] - **Turnover Rate**: 0.83X[42] - **Recent IC**: 0.0331*** (December 2025)[4][41] - **1-Year IC Mean**: 0.0669***[4][41] - **Annualized Return**: 37.40%[42] - **Annualized Volatility**: 23.39%[42] - **IR**: 1.60[42] - **Maximum Drawdown**: 27.29%[42] - **Annualized Excess Return**: 22.42%[4][42] --- Quantitative Factors and Construction Methods 1. Factor Name: GAN_GRU Factor - **Factor Construction Idea**: The GAN_GRU factor is derived from the GAN_GRU model, leveraging GAN for price-volume feature generation and GRU for time-series encoding[4][13][14] - **Factor Construction Process**: - The GAN generator processes raw price-volume time-series features (\( Input\_Shape = 40 \times 18 \)) and outputs transformed features with the same shape (\( Input\_Shape = 40 \times 18 \))[37] - The GRU component encodes these features into a predictive factor for stock selection[22] - The factor undergoes industry and market capitalization neutralization and standardization[22] - **Factor Evaluation**: The GAN_GRU factor demonstrates robust performance across various industries and time periods, with significant IC values and excess returns[4][41] --- Factor Backtesting Results 1. GAN_GRU Factor - **IC Mean**: 0.1119*** (2019-2025)[4][41] - **ICIR (non-annualized)**: 0.89[42] - **Turnover Rate**: 0.83X[42] - **Recent IC**: 0.0331*** (December 2025)[4][41] - **1-Year IC Mean**: 0.0669***[4][41] - **Annualized Return**: 37.40%[42] - **Annualized Volatility**: 23.39%[42] - **IR**: 1.60[42] - **Maximum Drawdown**: 27.29%[42] - **Annualized Excess Return**: 22.42%[4][42] 2. Industry-Specific Performance - **Top 5 Industries by Recent IC (October 2025)**: - Social Services: 0.4243*** - Coal: 0.2643*** - Environmental Protection: 0.2262*** - Retail: 0.1888*** - Steel: 0.1812***[4][41][42] - **Top 5 Industries by 1-Year IC Mean**: - Social Services: 0.1303*** - Steel: 0.1154*** - Non-Bank Financials: 0.1157*** - Retail: 0.1067*** - Building Materials: 0.1017***[4][41][42] 3. Industry-Specific Excess Returns - **Top 5 Industries by December 2025 Excess Returns**: - Banking: 4.30% - Real Estate: 3.51% - Environmental Protection: 2.18% - Retail: 1.76% - Machinery: 1.71%[2][45] - **Top 5 Industries by 1-Year Average Excess Returns**: - Banking: 2.12% - Real Estate: 1.93% - Environmental Protection: 1.50% - Retail: 1.46% - Machinery: 1.23%[2][46]

Core Viewpoint - The article discusses the recent changes in China's tariff schedule for 2026, particularly the introduction of new tariff categories for robots, including intelligent bionic robots and cleaning robots, aimed at supporting technological development and the circular economy [3][5]. Group 1: Tariff Adjustments - The 2026 tariff schedule includes new categories for intelligent bionic robots with a most-favored-nation (MFN) rate of 0% and a general rate of 30% [3][5]. - Two new categories for cleaning robots have been added, with MFN rates of 8% and 0%, and general rates of 130% and 30% respectively [3][5]. Group 2: Definition and Features of Intelligent Bionic Robots - Intelligent bionic robots are defined as autonomous or semi-autonomous robots that closely mimic the appearance, structure, or functions of humans or animals [4]. - These robots are equipped with various sensors and technologies, enabling them to perform tasks such as environmental perception, autonomous path planning, and human interaction through natural language processing [4]. Group 3: Industry Impact and Future Outlook - The addition of these tariff categories is expected to help industries and companies better understand trade data and assess overseas market trends [5]. - The tariff adjustments also include a temporary import tax rate below the MFN rate for 935 items, aimed at promoting high-level technological self-reliance and modern industrial system construction [5].

Core Insights - The Chinese government has introduced new tariff categories for intelligent bionic robots and cleaning robots as part of the 2026 tariff adjustment plan, with the most favored nation (MFN) rates set at 0% for intelligent bionic robots and varying rates for cleaning robots [1][3] Group 1: New Tariff Categories - The 2026 tariff adjustment plan includes the addition of intelligent bionic robot tariff categories with an MFN rate of 0% and a general rate of 30%, along with two cleaning robot categories with MFN rates of 8% and 0%, and general rates of 130% and 30% respectively [1][3] - Intelligent bionic robots are defined as autonomous or semi-autonomous robots that closely mimic the appearance, structure, or functions of humans or animals, equipped with advanced technologies for environmental perception and interaction [2] Group 2: Rationale for Changes - The adjustments aim to support technological advancement, circular economy, and the development of the forest economy, with the total number of tariff categories now reaching 8,972 [3] - The introduction of these new tariff categories is expected to help industries and companies accurately grasp trade data and assess overseas market trends [3]

来源：科技日报 科技日报记者 赵汉斌 记者29日从中国科学院云南天文台获悉，该台系外行星与太阳系小行星研究团组近期借助光谱巡天大数据与机器 学习技术，新发现了一大批磁活动年轻恒星，为研究恒星磁场起源、演化及行星形成机制提供了重要样本。相关 研究成果发表在国际期刊《天体物理学杂志增刊》上。 为做好研究准备，徐甫坤博士、顾盛宏研究员等人基于郭守敬望远镜巡天光谱数据库，通过降低光谱分辨率模拟 未来巡天空间望远镜主巡天光谱，开展预研究工作，并运用机器学习中的变分自编码器算法，对郭守敬望远镜/ 开普勒望远镜天区的巡天数据进行系统分析，通过诊断代表恒星磁场活动指标的氢阿尔法谱线发射强度以及代表 恒星年龄的锂谱线吸收强度，成功识别出500余颗磁活动年轻恒星，涵盖金牛座T型星、磁活动超饱和状态恒星等 多种类型。 研究团队还编制了新发现恒星的详细星表，系统测量其活动水平、年龄、周期，筛选出多个高价值目标源。据介 绍，后期对这些样本开展精细观测，将有助于揭示全对流恒星磁场活动特征与演化规律，为恒星内部磁场发电过 程提供关键观测约束，深化人类对恒星早期演化及行星形成中磁场作用的理解。 在恒星早期演化过程中，原恒星逐渐成形并吸积星周 ...