arXiv subjects

Yinuo Zhang

Publications and source records attributed to Yinuo Zhang.

33 records · Page 2Linked to original sources

arXiv2025

PepTune: De Novo Generation of Therapeutic Peptides with Multi-Objective-Guided Discrete Diffusion

We present PepTune, a multi-objective discrete diffusion model for simultaneous generation and optimization of therapeutic peptide SMILES. Built on the Masked Discrete Language Model (MDLM) framework, PepTune ensures valid peptide structures with a novel bond-dependent masking schedule and invalid loss function. To guide the diffusion process, we introduce Monte Carlo Tree Guidance (MCTG), an inference-time multi-objective guidance algorithm that balances exploration and exploitation to iteratively refine Pareto-optimal sequences. MCTG integrates classifier-based rewards with search-tree expansion, overcoming gradient estimation challenges and data sparsity. Using PepTune, we generate diverse, chemically-modified peptides simultaneously optimized for multiple therapeutic properties, including target binding affinity, membrane permeability, solubility, hemolysis, and non-fouling for various disease-relevant targets. In total, our results demonstrate that MCTG for masked discrete diffusion is a powerful and modular approach for multi-objective sequence design in discrete state spaces.

q-bio.BM↗

arXiv2025

Multi-Objective-Guided Discrete Flow Matching for Controllable Biological Sequence Design

Designing biological sequences that satisfy multiple, often conflicting, functional and biophysical criteria remains a central challenge in biomolecule engineering. While discrete flow matching models have recently shown promise for efficient sampling in high-dimensional sequence spaces, existing approaches address only single objectives or require continuous embeddings that can distort discrete distributions. We present Multi-Objective-Guided Discrete Flow Matching (MOG-DFM), a general framework to steer any pretrained discrete flow matching generator toward Pareto-efficient trade-offs across multiple scalar objectives. At each sampling step, MOG-DFM computes a hybrid rank-directional score for candidate transitions and applies an adaptive hypercone filter to enforce consistent multi-objective progression. We also trained two unconditional discrete flow matching models, PepDFM for diverse peptide generation and EnhancerDFM for functional enhancer DNA generation, as base generation models for MOG-DFM. We demonstrate MOG-DFM's effectiveness in generating peptide binders optimized across five properties (hemolysis, non-fouling, solubility, half-life, and binding affinity), and in designing DNA sequences with specific enhancer classes and DNA shapes. In total, MOG-DFM proves to be a powerful tool for multi-property-guided biomolecule sequence design.

cs.LG↗

arXiv2025

Write Cycling Endurance Exceeding 1010 in Sub-50 nm Ferroelectric AlScN

Wurtzite ferroelectrics, particularly aluminum scandium nitride (AlScN), have emerged as a promising materials platform for nonvolatile memories, offering high polarization values exceeding 100 uC/cm2. However, their high coercive fields (>3 MV/cm) have limited cycling endurance to ~107 cycles in previous reports. Here, we demonstrate unprecedented control of polarization switching in AlScN, achieving write cycling endurance exceeding 1010 cycles a thousand fold improvement over previous wurtzite ferroelectric benchmarks. Through precise voltage modulation in 45 nm thick Al0.64Sc0.36N capacitors, we show that while complete polarization reversal (2Pr ~ 200 uC/cm2) sustains ~108 cycles, partial switching extends endurance beyond 1010 cycles while maintaining a substantial polarization (>30 uC/cm2 for 2Pr). This exceptional endurance, combined with breakdown fields approaching 10 MV/cm in optimized 10 um diameter devices, represents the highest reported values for any wurtzite ferroelectric. Our findings establish a new paradigm for reliability in nitride ferroelectrics, demonstrating that controlled partial polarization and size scaling enables both high endurance and energy efficient operation.

cond-mat.mtrl-sci↗

arXiv2025

Low-voltage Ferroelectric Field-Effect Transistors with Ultrathin Aluminum Scandium Nitride and 2D channels

The continued evolution of CMOS technology demands materials and architectures that emphasize low power consumption, particularly for computations involving large scale data processing and multivariable optimization. Ferroelectric materials offer promising solutions through enabling dual-purpose memory units capable of performing both storage and logic operations. In this study, we demonstrate ferroelectric field effect transistors (FeFETs) with MoS2 monolayer channels fabricated on ultrathin 5 nm and 10 nm ferroelectric Aluminum Scandium Nitride (AlScN) films. By decreasing the thickness of the ferroelectric film, we achieve significantly reduced gate voltages (<3V) required to switch the conductance of the devices, enabling operation at low voltages compatible with advanced CMOS. We observe a characteristic crossover in hysteresis behavior that varies with film thickness, channel fabrication method, and environmental conditions. Through systematic investigation of multiple parameters including channel fabrication methods, dimensional scaling, and environmental effects, we provide pathways to improve device performance. While our devices demonstrate clear ferroelectric switching behavior, further optimization is required to enhance the ON/OFF ratio at zero gate voltage while continuing to reduce the coercive field of these ultrathin films.

physics.app-ph↗

arXiv2025

Observation of giant remnant polarization in ultrathin AlScN at cryogenic temperatures

The discovery of wurtzite ferroelectrics opens new frontiers in polar materials, yet their behavior at cryogenic temperatures remains unexplored. Here, we reveal unprecedented ferroelectric properties in ultrathin (10 nm) Al$_{0.68}$Sc$_{0.32}$N (AlScN) at cryogenic temperatures where the properties are fundamentally distinct from those of conventional oxide ferroelectrics. At 12 K, we demonstrate a giant remnant polarization exceeding 250 $μ$C/cm$^2$ -- more than twice that of any known ferroelectric -- driven by an enhanced c/a ratio in the wurtzite structure. Our devices sustain remarkably high electric fields (~13 MV/cm) while maintaining reliable switching, achieving over 104 polarization reversal cycles at 12 K. Critically, this breakdown field strength approaches that of passive dielectric materials while maintaining ferroelectric functionality. The extraordinary polarization enhancement and high-field stability at cryogenic temperatures contrasts sharply with oxide ferroelectrics, establishing wurtzite ferroelectrics as a distinct class of polar materials with implications spanning fundamental physics to cryogenic non-volatile memory and quantum technologies.

cond-mat.mtrl-sci↗

arXiv2025

Gumbel-Softmax Flow Matching with Straight-Through Guidance for Controllable Biological Sequence Generation

Flow matching in the continuous simplex has emerged as a promising strategy for DNA sequence design, but struggles to scale to higher simplex dimensions required for peptide and protein generation. We introduce Gumbel-Softmax Flow and Score Matching, a generative framework on the simplex based on a novel Gumbel-Softmax interpolant with a time-dependent temperature. Using this interpolant, we introduce Gumbel-Softmax Flow Matching by deriving a parameterized velocity field that transports from smooth categorical distributions to distributions concentrated at a single vertex of the simplex. We alternatively present Gumbel-Softmax Score Matching which learns to regress the gradient of the probability density. Our framework enables high-quality, diverse generation and scales efficiently to higher-dimensional simplices. To enable training-free guidance, we propose Straight-Through Guided Flows (STGFlow), a classifier-based guidance method that leverages straight-through estimators to steer the unconditional velocity field toward optimal vertices of the simplex. STGFlow enables efficient inference-time guidance using classifiers pre-trained on clean sequences, and can be used with any discrete flow method. Together, these components form a robust framework for controllable de novo sequence generation. We demonstrate state-of-the-art performance in conditional DNA promoter design, sequence-only protein generation, and target-binding peptide design for rare disease treatment.

cs.LG↗

arXiv2024

PepDoRA: A Unified Peptide Language Model via Weight-Decomposed Low-Rank Adaptation

Peptide therapeutics, including macrocycles, peptide inhibitors, and bioactive linear peptides, play a crucial role in therapeutic development due to their unique physicochemical properties. However, predicting these properties remains challenging. While structure-based models primarily focus on local interactions, language models are capable of capturing global therapeutic properties of both modified and linear peptides. Protein language models like ESM-2, though effective for natural peptides, cannot however encode chemical modifications. Conversely, pre-trained chemical language models excel in representing small molecule properties but are not optimized for peptides. To bridge this gap, we introduce PepDoRA, a unified peptide representation model. Leveraging Weight-Decomposed Low-Rank Adaptation (DoRA), PepDoRA efficiently fine-tunes the ChemBERTa-77M-MLM on a masked language model objective to generate optimized embeddings for downstream property prediction tasks involving both modified and unmodified peptides. By tuning on a diverse and experimentally valid set of 100,000 modified, bioactive, and binding peptides, we show that PepDoRA embeddings capture functional properties of input peptides, enabling the accurate prediction of membrane permeability, non-fouling and hemolysis propensity, and via contrastive learning, target protein-specific binding. Overall, by providing a unified representation for chemically and biologically diverse peptides, PepDoRA serves as a versatile tool for function and activity prediction, facilitating the development of peptide therapeutics across a broad spectrum of applications.

q-bio.BM↗

arXiv2024

A broadband hyperspectral image sensor with high spatio-temporal resolution

Hyperspectral imaging provides high-dimensional spatial-temporal-spectral information revealing intrinsic matter characteristics. Here we report an on-chip computational hyperspectral imaging framework with high spatial and temporal resolution. By integrating different broadband modulation materials on the image sensor chip, the target spectral information is non-uniformly and intrinsically coupled on each pixel with high light throughput. Using intelligent reconstruction algorithms, multi-channel images can be recovered from each frame, realizing real-time hyperspectral imaging. Following such a framework, we for the first time fabricated a broadband VIS-NIR (400-1700 nm) hyperspectral imaging sensor using photolithography, with an average light throughput of 74.8% and 96 wavelength channels. The demonstrated resolution is 1,024*1,024 pixels at 124 fps. We demonstrated its wide applications including chlorophyll and sugar quantification for intelligent agriculture, blood oxygen and water quality monitoring for human health, textile classification and apple bruise detection for industrial automation, and remote lunar detection for astronomy. The integrated hyperspectral image sensor weighs only tens of grams, and can be assembled on various resource-limited platforms or equipped with off-the-shelf optical systems. The technique transforms the challenge of high-dimensional imaging from a high-cost manufacturing and cumbersome system to one that is solvable through on-chip compression and agile computation.

physics.optics↗

arXiv2024

Scattering-induced entropy boost for highly-compressed optical sensing and encryption

Image sensing often relies on a high-quality machine vision system with a large field of view and high resolution. It requires fine imaging optics, has high computational costs, and requires a large communication bandwidth between image sensors and computing units. In this paper, we propose a novel image-free sensing framework for resource-efficient image classification, where the required number of measurements can be reduced by up to two orders of magnitude. In the proposed framework for single-pixel detection, the optical field for a target is first scattered by an optical diffuser and then two-dimensionally modulated by a spatial light modulator. The optical diffuser simultaneously serves as a compressor and an encryptor for the target information, effectively narrowing the field of view and improving the system's security. The one-dimensional sequence of intensity values, which is measured with time-varying patterns on the spatial light modulator, is then used to extract semantic information based on end-to-end deep learning. The proposed sensing framework is shown to obtain over a 95\% accuracy at sampling rates of 1% and 5% for classification on the MNIST dataset and the recognition of Chinese license plates, respectively, and the framework is up to 24% more efficient than the approach without an optical diffuser. The proposed framework represents a significant breakthrough in high-throughput machine intelligence for scene analysis with low bandwidth, low costs, and strong encryption.

cs.CV↗

arXiv2024

Suppressing seizure via optimal electrical stimulation to the hub of epileptic brain network

The electrical stimulation to the seizure onset zone (SOZ) serves as an efficient approach to seizure suppression. Recently, seizure dynamics have gained widespread attendance in its network propagation mechanisms. Compared with the direct stimulation to SOZ, other brain network-level approaches that can effectively suppress epileptic seizures remain under-explored. In this study, we introduce a platform equipped with a system identification module and a control strategy module, to validate the effectiveness of the hub of the epileptic brain network in suppressing seizure. The identified surrogate dynamics show high predictive performance in reconstructing neural dynamics which enables the model predictive framework to achieve accurate neural stimulation. The electrical stimulation on the hub of the epileptic brain network shows remarkable performance as the direct stimulation of SOZ in suppressing seizure dynamics. Underpinned by network control theory, our platform offers a general tool for the validation of neural stimulation.

q-bio.NC↗

arXiv2024

Reverse engineering the brain input: Network control theory to identify cognitive task-related control nodes

The human brain receives complex inputs when performing cognitive tasks, which range from external inputs via the senses to internal inputs from other brain regions. However, the explicit inputs to the brain during a cognitive task remain unclear. Here, we present an input identification framework for reverse engineering the control nodes and the corresponding inputs to the brain. The framework is verified with synthetic data generated by a predefined linear system, indicating it can robustly reconstruct data and recover the inputs. Then we apply the framework to the real motor-task fMRI data from 200 human subjects. Our results show that the model with sparse inputs can reconstruct neural dynamics in motor tasks ($EV=0.779$) and the identified 28 control nodes largely overlap with the motor system. Underpinned by network control theory, our framework offers a general tool for understanding brain inputs.

q-bio.NC↗

arXiv2024

Multi-State, Ultra-thin, BEOL-Compatible AlScN Ferroelectric Diodes

The growth in data generation necessitates efficient data processing technologies to address the von Neumann bottleneck in conventional computer architecture. Memory-driven computing, which integrates non-volatile memory (NVM) devices in a 3D stack, is gaining attention, with CMOS back-end-of-line (BEOL) compatible ferroelectric (FE) diodes being ideal due to their two-terminal design and inherently selector-free nature, facilitating high-density crossbar arrays. Here, we demonstrate BEOL-compatible, high-performance FE-diodes scaled to 5, 10, and 20 nm FE Al0.72Sc0.28N/Al0.64Sc0.36N films. Through interlayer (IL) engineering, we show substantial improvements in the ON/OFF ratios (>166 times) and rectification ratios (>176 times) in these scaled devices. The superlative characteristics also enables 5-bit multi-state operation with a stable retention. We also experimentally and theoretically demonstrate the counterintuitive result that the inclusion of an IL can lead to a decrease in the ferroelectric switching voltage of the device. An in-depth analysis into the device transport mechanisms is performed, and our compact model aligns seamlessly with the experimental results. Our results suggest the possibility of using scaled AlxSc1-xN FE-diodes for high performance, low-power, embedded NVM.

cond-mat.mtrl-sci↗

arXiv2023

Extreme Image Compression using Fine-tuned VQGANs

Recent advances in generative compression methods have demonstrated remarkable progress in enhancing the perceptual quality of compressed data, especially in scenarios with low bitrates. However, their efficacy and applicability to achieve extreme compression ratios ($<0.05$ bpp) remain constrained. In this work, we propose a simple yet effective coding framework by introducing vector quantization (VQ)--based generative models into the image compression domain. The main insight is that the codebook learned by the VQGAN model yields a strong expressive capacity, facilitating efficient compression of continuous information in the latent space while maintaining reconstruction quality. Specifically, an image can be represented as VQ-indices by finding the nearest codeword, which can be encoded using lossless compression methods into bitstreams. We propose clustering a pre-trained large-scale codebook into smaller codebooks through the K-means algorithm, yielding variable bitrates and different levels of reconstruction quality within the coding framework. Furthermore, we introduce a transformer to predict lost indices and restore images in unstable environments. Extensive qualitative and quantitative experiments on various benchmark datasets demonstrate that the proposed framework outperforms state-of-the-art codecs in terms of perceptual quality-oriented metrics and human perception at extremely low bitrates ($\le 0.04$ bpp). Remarkably, even with the loss of up to $20\%$ of indices, the images can be effectively restored with minimal perceptual loss.

cs.CV↗

arXiv2023

INO at Factify 2: Structure Coherence based Multi-Modal Fact Verification

This paper describes our approach to the multi-modal fact verification (FACTIFY) challenge at AAAI2023. In recent years, with the widespread use of social media, fake news can spread rapidly and negatively impact social security. Automatic claim verification becomes more and more crucial to combat fake news. In fact verification involving multiple modal data, there should be a structural coherence between claim and document. Therefore, we proposed a structure coherence-based multi-modal fact verification scheme to classify fake news. Our structure coherence includes the following four aspects: sentence length, vocabulary similarity, semantic similarity, and image similarity. Specifically, CLIP and Sentence BERT are combined to extract text features, and ResNet50 is used to extract image features. In addition, we also extract the length of the text as well as the lexical similarity. Then the features were concatenated and passed through the random forest classifier. Finally, our weighted average F1 score has reached 0.8079, achieving 2nd place in FACTIFY2.

cs.LG↗

arXiv2015

Shifted Hecke insertion and the K-theory of OG(n,2n+1)

Patrias and Pylyavskyy introduced shifted Hecke insertion as an application of their theory of dual filtered graphs. We use shifted Hecke insertion to construct symmetric function representatives for the K-theory of the orthogonal Grassmannian. These representatives are closely related to the shifted Grothendieck polynomials of Ikeda and Naruse. We then recover the K-theory structure coefficients of Clifford-Thomas-Yong/Buch-Samuel by introducing a shifted K-theoretic Poirier-Reutenauer algebra. Our proofs depend on the theory of shifted K-theoretic jeu de taquin and the weak K-Knuth relations.

math.CO↗