arXiv subjects

Qiang Cheng

Publications and source records attributed to Qiang Cheng.

At least 37 records · Page 2Linked to original sources

arXiv2025

Every Step Evolves: Scaling Reinforcement Learning for Trillion-Scale Thinking Model

We present Ring-1T, the first open-source, state-of-the-art thinking model with a trillion-scale parameter. It features 1 trillion total parameters and activates approximately 50 billion per token. Training such models at a trillion-parameter scale introduces unprecedented challenges, including train-inference misalignment, inefficiencies in rollout processing, and bottlenecks in the RL system. To address these, we pioneer three interconnected innovations: (1) IcePop stabilizes RL training via token-level discrepancy masking and clipping, resolving instability from training-inference mismatches; (2) C3PO++ improves resource utilization for long rollouts under a token budget by dynamically partitioning them, thereby obtaining high time efficiency; and (3) ASystem, a high-performance RL framework designed to overcome the systemic bottlenecks that impede trillion-parameter model training. Ring-1T delivers breakthrough results across critical benchmarks: 93.4 on AIME-2025, 86.72 on HMMT-2025, 2088 on CodeForces, and 55.94 on ARC-AGI-1. Notably, it attains a silver medal-level result on the IMO-2025, underscoring its exceptional reasoning capabilities. By releasing the complete 1T parameter MoE model to the community, we provide the research community with direct access to cutting-edge reasoning capabilities. This contribution marks a significant milestone in democratizing large-scale reasoning intelligence and establishes a new baseline for open-source model performance.

cs.CL↗

arXiv2025

Longwave-transparent low-emissivity material

Low emissivity (low-e) materials are crucial for conserving thermal energy in buildings, cold chain logistics and transportation by minimizing unwanted radiative heat loss or gain. However, their metallic nature intrinsically causes severe longwave attenuation, hindering their broad applications. Here, we introduce, for the first time, an all-dielectric longwave-transparent low-emissivity material (LLM) with ultra-broadband, high transmittance spanning 9 orders of magnitude, from terahertz to kilohertz frequencies. This meter-scale LLM not only achieves energy savings of up to 41.1% over commercial white paint and 10.2% over traditional low-e materials, but also unlocks various fundamentally new capabilities including high-speed wireless communication in energy-efficient buildings, wireless energy transfer with radiative thermal insulation, as well as non-invasive terahertz security screening and radio frequency identification in cold chain logistics. Our approach represents a new photonic solution towards carbon neutrality and smart city development, paving the way for a more sustainable and interconnected future.

physics.optics↗

arXiv2025

CrunchLLM: Multitask LLMs for Structured Business Reasoning and Outcome Prediction

Predicting the success of start-up companies, defined as achieving an exit through acquisition or IPO, is a critical problem in entrepreneurship and innovation research. Datasets such as Crunchbase provide both structured information (e.g., funding rounds, industries, investor networks) and unstructured text (e.g., company descriptions), but effectively leveraging this heterogeneous data for prediction remains challenging. Traditional machine learning approaches often rely only on structured features and achieve moderate accuracy, while large language models (LLMs) offer rich reasoning abilities but struggle to adapt directly to domain-specific business data. We present \textbf{CrunchLLM}, a domain-adapted LLM framework for startup success prediction. CrunchLLM integrates structured company attributes with unstructured textual narratives and applies parameter-efficient fine-tuning strategies alongside prompt optimization to specialize foundation models for entrepreneurship data. Our approach achieves accuracy exceeding 80\% on Crunchbase startup success prediction, significantly outperforming traditional classifiers and baseline LLMs. Beyond predictive performance, CrunchLLM provides interpretable reasoning traces that justify its predictions, enhancing transparency and trustworthiness for financial and policy decision makers. This work demonstrates how adapting LLMs with domain-aware fine-tuning and structured--unstructured data fusion can advance predictive modeling of entrepreneurial outcomes. CrunchLLM contributes a methodological framework and a practical tool for data-driven decision making in venture capital and innovation policy.

cs.LG↗

arXiv2025

Equal-spin and oblique-spin crossed Andreev reflections in ferromagnet/Ising superconductor/ferromagnet junction

We study the subgap transport through a ferromagnet/Ising superconductor/ferromagnet (F/ISC/F) junction by solving the Bogoliubov-de Gennes equations. It is found that the crossed Andreev reflection (CAR) and local Andreev reflection (LAR) strongly depend on the spin polarized F, the magnetization direction, and the Ising superconducting phase. For the same magnetization directions of the two F leads, the equal-spin CAR could take place due to spin-flip mechanism induced by the Ising spin-orbit coupling and equal-spin-triplet pairing. Both equal-spin CAR and equal-spin LAR exhibit a remarkable magnetoanisotropy with period πand show oscillatory behavior with chemical potential. The equal-spin CAR is more prominent for half-metal F and double-band ISC while the normal CAR is completely suppressed. When the magnetization directions of the two F leads are different, the oblique-spin CAR occurs and its magnetoanisotropic period generally becomes 2πinstead of π. In the oblique-spin CAR process, the spins of the electron and hole are neither parallel nor antiparallel. Furthermore, the property of oblique-spin CAR is very sensitive to the spin and valley degrees of freedom. The spin and valley polarized CAR can be achieved and controlled by the chemical potentials and the magnetization directions.

cond-mat.supr-con↗

arXiv2025

Controllable Josephson diode effect, $0$-$π$ transition and switch effect in the superconductor/two-dimensional Weyl nodal line semimetal/superconductor junctions

We study the Josephson effects in the superconductor/two-dimensional Weyl nodal line semimetal/superconductor junctions using the Green's function method. When the Rashba spin-orbit coupling and an external magnetic field coexist in the semimetal, the symmetries protecting the reciprocity of supercurrent can be broken and the Josephson diode effect with the nonreciprocity of supercurrent can be realized. A high efficiency exceeding $40\%$ can be achieved with the experimentally accessible values of the magnetic field and the Rashba spin-orbit coupling. The diode efficiency can be easily controlled by the direction and magnitude of the field and the strength of the spin-orbit coupling. When the spin-orbit coupling is absent or the external field is absent, the Josephson diode effect vanishes but the current-phase difference relations still show strong dependence on the field or the coupling. The tunable $0$-$π$ transition and the switch effect of supercurrent in the junctions can be formed if the direction of the field is rotated or the magnitude of the field and the strength of the coupling are changed. The obtained Josephson diode effect, the $0$-$π$ transition and the switch effect of supercurrent are helpful in the design of the quantum devices based on nodal line semimetals.

cond-mat.supr-con↗

arXiv2025

MolSnap: Snap-Fast Molecular Generation with Latent Variational Mean Flow

Molecular generation conditioned on textual descriptions is a fundamental task in computational chemistry and drug discovery. Existing methods often struggle to simultaneously ensure high-quality, diverse generation and fast inference. In this work, we propose a novel causality-aware framework that addresses these challenges through two key innovations. First, we introduce a Causality-Aware Transformer (CAT) that jointly encodes molecular graph tokens and text instructions while enforcing causal dependencies during generation. Second, we develop a Variational Mean Flow (VMF) framework that generalizes existing flow-based methods by modeling the latent space as a mixture of Gaussians, enhancing expressiveness beyond unimodal priors. VMF enables efficient one-step inference while maintaining strong generation quality and diversity. Extensive experiments on four standard molecular benchmarks demonstrate that our model outperforms state-of-the-art baselines, achieving higher novelty (up to 74.5\%), diversity (up to 70.3\%), and 100\% validity across all datasets. Moreover, VMF requires only one number of function evaluation (NFE) during conditional generation and up to five NFEs for unconditional generation, offering substantial computational efficiency over diffusion-based methods.

cs.LG↗

arXiv2025

DepMicroDiff: Diffusion-Based Dependency-Aware Multimodal Imputation for Microbiome Data

Microbiome data analysis is essential for understanding host health and disease, yet its inherent sparsity and noise pose major challenges for accurate imputation, hindering downstream tasks such as biomarker discovery. Existing imputation methods, including recent diffusion-based models, often fail to capture the complex interdependencies between microbial taxa and overlook contextual metadata that can inform imputation. We introduce DepMicroDiff, a novel framework that combines diffusion-based generative modeling with a Dependency-Aware Transformer (DAT) to explicitly capture both mutual pairwise dependencies and autoregressive relationships. DepMicroDiff is further enhanced by VAE-based pretraining across diverse cancer datasets and conditioning on patient metadata encoded via a large language model (LLM). Experiments on TCGA microbiome datasets show that DepMicroDiff substantially outperforms state-of-the-art baselines, achieving higher Pearson correlation (up to 0.712), cosine similarity (up to 0.812), and lower RMSE and MAE across multiple cancer types, demonstrating its robustness and generalizability for microbiome imputation.

cs.LG↗

arXiv2025

Five-Gene Expression Formula Accurately Detects Hepatocellular Carcinoma Tumors

Hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related deaths worldwide. Several diagnostic methods, such as imaging modalities and Serum Alpha-Fetoprotein (AFP) testing, have been used for HCC detection; however, their effectiveness is limited to later stages of the disease. In contrast, transcriptomic analysis of biposy samples has shown promise for early detection. While machine learning techniques have been applied to transcriptomic data for cancer detection, their clinical adoption remains limited due to challenges such as poor generalizability across different datasets, lack of interpretability, and high computational complexity. To address these limitations, we developed a novel predictive formula for HCC detection using the Kolmogorov-Arnold Network (KAN). This formula is based on the expression levels of five genes: VIPR1, CYP1A2, FCN3, ECM1, and LIFR. Derived from the GSE25097 dataset, the formula offers a simple, interpretable, efficient, and accessible approach for HCC identification. It achieves 99% accuracy on the GSE25097 test set and demonstrates robust performance on six additional independent datasets, achieving accuracies of above 90% in all cases. These findings highlight the critical role of these five genes as biomarkers for HCC detection, offering a foundation for future research and clinical applications to improve HCC diagnostic approaches.

q-bio.GN↗

arXiv2025

Topology Design of Reconfigurable Intelligent Surfaces Based on Current Distribution and Otsu Image Segmentation

Miniaturization of reconffgurable intelligent surface RIS) elements is a crucial trend in the development of RISs. It not only facilitates the attainment of multifunctional integration but also promotes seamless amalgamation with other elements. The current on the RIS element plays a crucial role in determining the characteristics of the induced electromagnetic ffeld components. Segments with high current intensity determine the performance of RIS elements. Carving the parts with strong current distribution density into the metal patch of RIS element structure can achieve miniaturization. Based on this insight, this work proposes a topology design method that leverages current distribution and image processing techniques to achieve efffcient miniaturization of the RIS elements. In this proposed method, we ffrst obtain the current distribution across different operational states and the period of the working frequency. Next, we employ the Otsu image segmentation method to extract relevant image information from the current distribution images of the RIS elements. Subsequently, we utilize linear mapping techniques to convert this image information into the structure of RIS elements. Then, based on the structure of the RIS elements, the Quasi-Newton optimization algorithm is utilized to obtain the parameters of the tunable device that correspond to various operational states. As a result, we successfully construct the structural topology of the RIS elements based on their current distribution, designing areas with strong current distribution as metal patches. To validate the performance of the proposed method, a 16 by 16 3-bit RIS was developed, fabricated and measured. Compared with existing RIS designs, the proportion of the top-layer metal patches is smaller, which provides the possibility for integrating other functions and devices.

physics.app-ph↗

arXiv2025

Evaluating Deep Learning Models for Breast Cancer Classification: A Comparative Study

This study evaluates the effectiveness of deep learning models in classifying histopathological images for early and accurate detection of breast cancer. Eight advanced models, including ResNet-50, DenseNet-121, ResNeXt-50, Vision Transformer (ViT), GoogLeNet (Inception v3), EfficientNet, MobileNet, and SqueezeNet, were compared using a dataset of 277,524 image patches. The Vision Transformer (ViT) model, with its attention-based mechanisms, achieved the highest validation accuracy of 94%, outperforming conventional CNNs. The study demonstrates the potential of advanced machine learning methods to enhance precision and efficiency in breast cancer diagnosis in clinical settings.

eess.IV↗

arXiv2025

TSCMamba: Mamba Meets Multi-View Learning for Time Series Classification

Multivariate time series classification (TSC) is critical for various applications in fields such as healthcare and finance. While various approaches for TSC have been explored, important properties of time series, such as shift equivariance and inversion invariance, are largely underexplored by existing works. To fill this gap, we propose a novel multi-view approach to capture patterns with properties like shift equivariance. Our method integrates diverse features, including spectral, temporal, local, and global features, to obtain rich, complementary contexts for TSC. We use continuous wavelet transform to capture time-frequency features that remain consistent even when the input is shifted in time. These features are fused with temporal convolutional or multilayer perceptron features to provide complex local and global contextual information. We utilize the Mamba state space model for efficient and scalable sequence modeling and capturing long-range dependencies in time series. Moreover, we introduce a new scanning scheme for Mamba, called tango scanning, to effectively model sequence relationships and leverage inversion invariance, thereby enhancing our model's generalization and robustness. Experiments on two sets of benchmark datasets (10+20 datasets) demonstrate our approach's effectiveness, achieving average accuracy improvements of 4.01-6.45\% and 7.93\% respectively, over leading TSC models such as TimesNet and TSLANet.

cs.LG↗

arXiv2025

Mol-CADiff: Causality-Aware Autoregressive Diffusion for Molecule Generation

The design of novel molecules with desired properties is a key challenge in drug discovery and materials science. Traditional methods rely on trial-and-error, while recent deep learning approaches have accelerated molecular generation. However, existing models struggle with generating molecules based on specific textual descriptions. We introduce Mol-CADiff, a novel diffusion-based framework that uses causal attention mechanisms for text-conditional molecular generation. Our approach explicitly models the causal relationship between textual prompts and molecular structures, overcoming key limitations in existing methods. We enhance dependency modeling both within and across modalities, enabling precise control over the generation process. Our extensive experiments demonstrate that Mol-CADiff outperforms state-of-the-art methods in generating diverse, novel, and chemically valid molecules, with better alignment to specified properties, enabling more intuitive language-driven molecular design.

cs.LG↗

arXiv2025

CausalGeD: Blending Causality and Diffusion for Spatial Gene Expression Generation

The integration of single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) data is crucial for understanding gene expression in spatial context. Existing methods for such integration have limited performance, with structural similarity often below 60\%, We attribute this limitation to the failure to consider causal relationships between genes. We present CausalGeD, which combines diffusion and autoregressive processes to leverage these relationships. By generalizing the Causal Attention Transformer from image generation to gene expression data, our model captures regulatory mechanisms without predefined relationships. Across 10 tissue datasets, CausalGeD outperformed state-of-the-art baselines by 5- 32\% in key metrics, including Pearson's correlation and structural similarity, advancing both technical and biological insights.

cs.CV↗

arXiv2025

GraphMinNet: Learning Dependencies in Graphs with Light Complexity Minimal Architecture

Graph Neural Networks (GNNs) have demonstrated remarkable success in various applications, yet they often struggle to capture long-range dependencies (LRD) effectively. This paper introduces GraphMinNet, a novel GNN architecture that generalizes the idea of minimal Gated Recurrent Units to graph-structured data. Our approach achieves efficient LRD modeling with linear computational complexity while maintaining permutation equivariance and stability. The model incorporates both structural and positional information through a unique combination of feature and positional encodings, leading to provably stronger expressiveness than the 1-WL test. Theoretical analysis establishes that GraphMinNet maintains non-decaying gradients over long distances, ensuring effective long-range information propagation. Extensive experiments on ten diverse datasets, including molecular graphs, image graphs, and synthetic networks, demonstrate that GraphMinNet achieves state-of-the-art performance while being computationally efficient. Our results show superior performance on 6 out of 10 datasets and competitive results on the others, validating the effectiveness of our approach in capturing both local and global graph structures.

cs.LG↗

arXiv2025

Physics-Informed Machine Learning for Efficient Reconfigurable Intelligent Surface Design

Reconfigurable intelligent surface (RIS) is a two-dimensional periodic structure integrated with a large number of reflective elements, which can manipulate electromagnetic waves in a digital way, offering great potentials for wireless communication and radar detection applications. However, conventional RIS designs highly rely on extensive full-wave EM simulations that are extremely time-consuming. To address this challenge, we propose a machine-learning-assisted approach for efficient RIS design. An accurate and fast model to predict the reflection coefficient of RIS element is developed by combining a multi-layer perceptron neural network (MLP) and a dual-port network, which can significantly reduce tedious EM simulations in the network training. A RIS has been practically designed based on the proposed method. To verify the proposed method, the RIS has also been fabricated and measured. The experimental results are in good agreement with the simulation results, which validates the efficacy of the proposed method in RIS design.

cs.LG↗

arXiv2025

PROTECT: Protein circadian time prediction using unsupervised learning

Circadian rhythms regulate the physiology and behavior of humans and animals. Despite advancements in understanding these rhythms and predicting circadian phases at the transcriptional level, predicting circadian phases from proteomic data remains elusive. This challenge is largely due to the scarcity of time labels in proteomic datasets, which are often characterized by small sample sizes, high dimensionality, and significant noise. Furthermore, existing methods for predicting circadian phases from transcriptomic data typically rely on prior knowledge of known rhythmic genes, making them unsuitable for proteomic datasets. To address this gap, we developed a novel computational method using unsupervised deep learning techniques to predict circadian sample phases from proteomic data without requiring time labels or prior knowledge of proteins or genes. Our model involves a two-stage training process optimized for robust circadian phase prediction: an initial greedy one-layer-at-a-time pre-training which generates informative initial parameters followed by fine-tuning. During fine-tuning, a specialized loss function guides the model to align protein expression levels with circadian patterns, enabling it to accurately capture the underlying rhythmic structure within the data. We tested our method on both time-labeled and unlabeled proteomic data. For labeled data, we compared our predictions to the known time labels, achieving high accuracy, while for unlabeled human datasets, including postmortem brain regions and urine samples, we explored circadian disruptions. Notably, our analysis identified disruptions in rhythmic proteins between Alzheimer's disease and control subjects across these samples.

cs.LG↗

arXiv2024

A wideband amplifying and filtering reconfigurable intelligent surface for wireless relay

Programmable metasurfaces have garnered significant attention due to their exceptional ability to manipulate electromagnetic (EM) waves in real time, leading to the emergence of a prominent area in wireless communication, namely reconfigurable intelligent surfaces (RISs), to control the signal propagation and coverage. However, the existing RISs usually suffer from limited operating distance and band interference, which hinder their practical applications in wireless relay and communication systems. To overcome the limitations, we propose an amplifying and filtering RIS (AF-RIS) to enhance the in-band signal energy and filter the out-of-band signal of the incident EM waves, ensuring the miniaturization of the RIS array and enabling its anti-interference ability. In addition, each AF-RIS element is equipped with a 2-bit phase control capability, further endowing the entire array with great beamforming performance. An elaborately designed 4*8 AF-RIS array is presented by integrating the power dividing and combining networks, which substantially reduces the number of amplifiers and filters, thereby reducing the hardware costs and power consumption. Experimental results showcase the powerful capabilities of AF-RIS in beam-steering, frequency selectivity, and signal amplification. Therefore, the proposed AF-RIS holds significant promise for critical applications in wireless relay systems by offering an efficient solution to improve frequency selectivity, enhance signal coverage, and reduce hardware size.

eess.SP↗

arXiv2024

Deep Learning in Single-Cell and Spatial Transcriptomics Data Analysis: Advances and Challenges from a Data Science Perspective

The development of single-cell and spatial transcriptomics has revolutionized our capacity to investigate cellular properties, functions, and interactions in both cellular and spatial contexts. However, the analysis of single-cell and spatial omics data remains challenging. First, single-cell sequencing data are high-dimensional and sparse, often contaminated by noise and uncertainty, obscuring the underlying biological signals. Second, these data often encompass multiple modalities, including gene expression, epigenetic modifications, and spatial locations. Integrating these diverse data modalities is crucial for enhancing prediction accuracy and biological interpretability. Third, while the scale of single-cell sequencing has expanded to millions of cells, high-quality annotated datasets are still limited. Fourth, the complex correlations of biological tissues make it difficult to accurately reconstruct cellular states and spatial contexts. Traditional feature engineering-based analysis methods struggle to deal with the various challenges presented by intricate biological networks. Deep learning has emerged as a powerful tool capable of handling high-dimensional complex data and automatically identifying meaningful patterns, offering significant promise in addressing these challenges. This review systematically analyzes these challenges and discusses related deep learning approaches. Moreover, we have curated 21 datasets from 9 benchmarks, encompassing 58 computational methods, and evaluated their performance on the respective modeling tasks. Finally, we highlight three areas for future development from a technical, dataset, and application perspective. This work will serve as a valuable resource for understanding how deep learning can be effectively utilized in single-cell and spatial transcriptomics analyses, while inspiring novel approaches to address emerging challenges.

q-bio.GN↗