Yimin Liu

Xiangyi Li, Wenbo Chen, Yimin Liu et al. · berkeley

Agent Skills are structured packages of procedural knowledge that augment LLM agents at inference time. Despite rapid adoption, there is no standard way to measure whether they actually help. We present SkillsBench, a benchmark of 86 tasks across 11 domains paired with curated Skills and deterministic verifiers. Each task is evaluated under three conditions: no Skills, curated Skills, and self-generated Skills. We test 7 agent-model configurations over 7,308 trajectories. Curated Skills raise average pass rate by 16.2 percentage points(pp), but effects vary widely by domain (+4.5pp for Software Engineering to +51.9pp for Healthcare) and 16 of 84 tasks show negative deltas. Self-generated Skills provide no benefit on average, showing that models cannot reliably author the procedural knowledge they benefit from consuming. Focused Skills with 2--3 modules outperform comprehensive documentation, and smaller models with Skills can match larger models without them.

Xiangyi Li, Kyoung Whan Choe, Yimin Liu et al. · apple-ml

Large language model (LLM) agents are increasingly deployed to automate productivity tasks (e.g., email, scheduling, document management), but evaluating them on live services is risky due to potentially irreversible changes. Existing benchmarks rely on simplified environments and fail to capture realistic, stateful, multi-service workflows. We introduce ClawsBench, a benchmark for evaluating and improving LLM agents in realistic productivity settings. It includes five high-fidelity mock services (Gmail, Slack, Google Calendar, Google Docs, Google Drive) with full state management and deterministic snapshot/restore, along with 44 structured tasks covering single-service, cross-service, and safety-critical scenarios. We decompose agent scaffolding into two independent levers (domain skills that inject API knowledge via progressive disclosure, and a meta prompt that coordinates behavior across services) and vary both to measure their separate and combined effects. Experiments across 6 models, 4 agent harnesses, and 33 conditions show that with full scaffolding, agents achieve task success rates of 39-64% but exhibit unsafe action rates of 7-33%. On OpenClaw, the top five models fall within a 10 percentage-point band on task success (53-63%), with unsafe action rates from 7% to 23% and no consistent ordering between the two metrics. We identify eight recurring patterns of unsafe behavior, including multi-step sandbox escalation and silent contract modification.

Jie Yuan, Lei Wang, Yanhao Wang et al.

This paper introduces a robust discrimination method for distinguishing real ship targets from corner-reflector-array jamming with frequency-agile radar. The key idea is to exploit the multidimensional micro-motion signatures that separate rigid ships from non-rigid decoys. From Range-Velocity maps we derive two new hand-crafted descriptors-mean weighted residual (MWR) and complementary contrast factor (CCF) and fuse them with deep features learned by a lightweight CNN. An XGBoost classifier then gives the final decision. Extensive simulations show that the hybrid feature set consistently outperforms state-of-the-art alternatives, confirming the superiority of the proposed approach.

Shuting Jin, Yi Jiang, Yimin Liu et al.

Background: Identifying new indications for approved drugs is a complex and time-consuming process that requires extensive knowledge of pharmacology, clinical data, and advanced computational methods. Recently, deep learning (DL) methods have shown their capability for the accurate prediction of drug repositioning. However, implementing DL-based modeling requires in-depth domain knowledge and proficient programming skills. Results: In this application, we introduce DeepDR, the first integrated platform that combines a variety of established DL-based models for disease- and target-specific drug repositioning tasks. DeepDR leverages invaluable experience to recommend candidate drugs, which covers more than 15 networks and a comprehensive knowledge graph that includes 5.9 million edges across 107 types of relationships connecting drugs, diseases, proteins/genes, pathways, and expression from six existing databases and a large scientific corpus of 24 million PubMed publications. Additionally, the recommended results include detailed descriptions of the recommended drugs and visualize key patterns with interpretability through a knowledge graph. Conclusion: DeepDR is free and open to all users without the requirement of registration. We believe it can provide an easy-to-use, systematic, highly accurate, and computationally automated platform for both experimental and computational scientists.

Shlok Mishra, Tsung-Yu Lin, Linda Wang et al.

We present Xray-Visual, a unified vision model architecture for large-scale image and video understanding trained on industry-scale social media data. Our model leverages over 15 billion curated image-text pairs and 10 billion video-hashtag pairs from Facebook and Instagram, employing robust data curation pipelines that incorporate balancing and noise suppression strategies to maximize semantic diversity while minimizing label noise. We introduce a three-stage training pipeline that combines self-supervised MAE, semi-supervised hashtag classification, and CLIP-style contrastive learning to jointly optimize image and video modalities. Our architecture builds on a Vision Transformer backbone enhanced with efficient token reorganization (EViT) for improved computational efficiency. Extensive experiments demonstrate that Xray-Visual achieves state-of-the-art performance across diverse benchmarks, including ImageNet for image classification, Kinetics and HMDB51 for video understanding, and MSCOCO for cross-modal retrieval. The model exhibits strong robustness to domain shift and adversarial perturbations. We further demonstrate that integrating large language models as text encoders (LLM2CLIP) significantly enhances retrieval performance and generalization capabilities, particularly in real-world environments. Xray-Visual establishes new benchmarks for scalable, multimodal vision models, while maintaining superior accuracy and computational efficiency.

Yunliang Qi, Meng Lou, Yimin Liu et al.

Remote sensing image super-resolution (RSISR) is a crucial task in remote sensing image processing, aiming to reconstruct high-resolution (HR) images from their low-resolution (LR) counterparts. Despite the growing number of RSISR methods proposed in recent years, a systematic and comprehensive review of these methods is still lacking. This paper presents a thorough review of RSISR algorithms, covering methodologies, datasets, and evaluation metrics. We provide an in-depth analysis of RSISR methods, categorizing them into supervised, unsupervised, and quality evaluation approaches, to help researchers understand current trends and challenges. Our review also discusses the strengths, limitations, and inherent challenges of these techniques. Notably, our analysis reveals significant limitations in existing methods, particularly in preserving fine-grained textures and geometric structures under large-scale degradation. Based on these findings, we outline future research directions, highlighting the need for domain-specific architectures and robust evaluation protocols to bridge the gap between synthetic and real-world RSISR scenarios.

Guozheng Sun, Lei Wang, Yanhao Wang et al.

Radar automatic target recognition (RATR) based on high-resolution range profile (HRRP) has attracted increasing attention due to its ability to capture fine-grained structural features. However, recognizing targets under electronic countermeasures (ECM), especially the mainstream interrupted-sampling repeater jamming (ISRJ), remains a significant challenge, as HRRPs often suffer from serious feature distortion. To address this, we propose a robust HRRP recognition method guided by prior jamming information. Specifically, we introduce a point spread function (PSF) as prior information to model the HRRP distortion induced by ISRJ. Based on this, we design a recognition network that leverages this prior through a prior-guided feature interaction module and a hybrid loss function to enhance the model's discriminative capability. With the aid of prior information, the model can learn invariant features within distorted HRRP under different jamming parameters. Both the simulated and measured-data experiments demonstrate that our method consistently outperforms state-of-the-art approaches and exhibits stronger generalization capabilities when facing unseen jamming parameters.

Yongle Zhang, Yimin Liu, Yan Huang et al.

Text-guided diffusion models have achieved remarkable success in object inpainting by providing high-level semantic guidance through text prompts. However, they often lack precise pixel-level spatial control, especially in scenarios involving partially corrupted objects where critical uncorrupted cues remain. To overcome this limitation, sketch-guided methods have been introduced, using either indirect gradient modulation or direct sketch injection to improve structural control. Yet, existing approaches typically establish a one-way mapping from the sketch to the masked regions only, neglecting the contextual information from unmasked object areas. This leads to a disconnection between the sketch and the uncorrupted content, thereby causing sketch-guided inconsistency and structural mismatch. To tackle this challenge, we propose a sketch-guided bidirectional feature interaction framework built upon a pretrained Stable Diffusion model. Our bidirectional interaction features two complementary directions, context-to-sketch and sketch-to-inpainting, that enable fine-grained spatial control for partially corrupted object inpainting. In the context-to-sketch direction, multi-scale latents from uncorrupted object regions are propagated to the sketch branch to generate a visual mask that adapts the sketch features to the visible context and denoising progress. In the sketch-to-inpainting direction, a sketch-conditional affine transformation modulates the influence of sketch guidance based on the learned visual mask, ensuring consistency with uncorrupted object content. This interaction is applied at multiple scales within the encoder of the diffusion U-Net, enabling the model to restore object structures with enhanced spatial fidelity. Extensive experiments on two newly constructed benchmark datasets demonstrate that our approach outperforms state-of-the-art methods.

Zhaoyang Larry Jin, Mehdi Maasoumy, Yimin Liu et al.

Today's global supply chains face growing challenges due to rapidly changing market conditions, increased network complexity and inter-dependency, and dynamic uncertainties in supply, demand, and other factors. To combat these challenges, organizations employ Material Requirements Planning (MRP) software solutions to set inventory stock buffers - for raw materials, work-in-process goods, and finished products - to help them meet customer service levels. However, holding excess inventory further complicates operations and can lock up millions of dollars of capital that could be otherwise deployed. Furthermore, most commercially available MRP solutions fall short in considering uncertainties and do not result in optimal solutions for modern enterprises. At C3 AI, we fundamentally reformulate the inventory management problem as a constrained stochastic optimization. We then propose a simulation-optimization framework that minimizes inventory and related costs while maintaining desired service levels. The framework's goal is to find the optimal reorder parameters that minimize costs subject to a pre-defined service-level constraint and all other real-world operational constraints. These optimal reorder parameters can be fed back into an MRP system to drive optimal order placement, or used to place optimal orders directly. This approach has proven successful in reducing inventory levels by 10-35 percent, resulting in hundreds of millions of dollars of economic benefit for major enterprises at a global scale.

Yuwei Cheng, Mengxin Jiang, Jiannan Zhu et al.

Unmanned surface vehicles (USVs) have great value with their ability to execute hazardous and time-consuming missions over water surfaces. Recently, USVs for inland waterways have attracted increasing attention for their potential application in autonomous monitoring, transportation, and cleaning. However, unlike sailing in open water, the challenges posed by scenes of inland waterways, such as the complex distribution of obstacles, the global positioning system (GPS) signal denial environment, the reflection of bank-side structures, and the fog over the water surface, all impede USV application in inland waterways. To address these problems and stimulate relevant research, we introduce USVInland, a multisensor dataset for USVs in inland waterways. The collection of USVInland spans a trajectory of more than 26 km in diverse real-world scenes of inland waterways using various modalities, including lidar, stereo cameras, millimeter-wave radar, GPS, and inertial measurement units (IMUs). Based on the requirements and challenges in the perception and navigation of USVs for inland waterways, we build benchmarks for simultaneous localization and mapping (SLAM), stereo matching, and water segmentation. We evaluate common algorithms for the above tasks to determine the influence of unique inland waterway scenes on algorithm performance. Our dataset and the development tools are available online at https://www.orca-tech.cn/datasets.html.

Yimin Liu, Louis J. Durlofsky

Geological parameterization enables the representation of geomodels in terms of a relatively small set of variables. Parameterization is therefore very useful in the context of data assimilation and uncertainty quantification. In this study, a deep-learning-based geological parameterization algorithm, CNN-PCA, is developed for complex 3D geomodels. CNN-PCA entails the use of convolutional neural networks as a post-processor for the low-dimensional principal component analysis representation of a geomodel. The 3D treatments presented here differ somewhat from those used in the 2D CNN-PCA procedure. Specifically, we introduce a new supervised-learning-based reconstruction loss, which is used in combination with style loss and hard data loss. The style loss uses features extracted from a 3D CNN pretrained for video classification. The 3D CNN-PCA algorithm is applied for the generation of conditional 3D realizations, defined on $60\times60\times40$ grids, for three geological scenarios (binary and bimodal channelized systems, and a three-facies channel-levee-mud system). CNN-PCA realizations are shown to exhibit geological features that are visually consistent with reference models generated using object-based methods. Statistics of flow responses ($\text{P}_{10}$, $\text{P}_{50}$, $\text{P}_{90}$ percentile results) for test sets of 3D CNN-PCA models are shown to be in consistent agreement with those from reference geomodels. Lastly, CNN-PCA is successfully applied for history matching with ESMDA for the bimodal channelized system.

Meng Tang, Yimin Liu, Louis J. Durlofsky

A deep-learning-based surrogate model is developed and applied for predicting dynamic subsurface flow in channelized geological models. The surrogate model is based on deep convolutional and recurrent neural network architectures, specifically a residual U-Net and a convolutional long short term memory recurrent network. Training samples entail global pressure and saturation maps, at a series of time steps, generated by simulating oil-water flow in many (1500 in our case) realizations of a 2D channelized system. After training, the `recurrent R-U-Net' surrogate model is shown to be capable of accurately predicting dynamic pressure and saturation maps and well rates (e.g., time-varying oil and water rates at production wells) for new geological realizations. Assessments demonstrating high surrogate-model accuracy are presented for an individual geological realization and for an ensemble of 500 test geomodels. The surrogate model is then used for the challenging problem of data assimilation (history matching) in a channelized system. For this study, posterior reservoir models are generated using the randomized maximum likelihood method, with the permeability field represented using the recently developed CNN-PCA parameterization. The flow responses required during the data assimilation procedure are provided by the recurrent R-U-Net. The overall approach is shown to lead to substantial reduction in prediction uncertainty. High-fidelity numerical simulation results for the posterior geomodels (generated by the surrogate-based data assimilation procedure) are shown to be in essential agreement with the recurrent R-U-Net predictions. The accuracy and dramatic speedup provided by the surrogate model suggest that it may eventually enable the application of more formal posterior sampling methods in realistic problems.

Yimin Liu, Wenyue Sun, Louis J. Durlofsky

A new low-dimensional parameterization based on principal component analysis (PCA) and convolutional neural networks (CNN) is developed to represent complex geological models. The CNN-PCA method is inspired by recent developments in computer vision using deep learning. CNN-PCA can be viewed as a generalization of an existing optimization-based PCA (O-PCA) method. Both CNN-PCA and O-PCA entail post-processing a PCA model to better honor complex geological features. In CNN-PCA, rather than use a histogram-based regularization as in O-PCA, a new regularization involving a set of metrics for multipoint statistics is introduced. The metrics are based on summary statistics of the nonlinear filter responses of geological models to a pre-trained deep CNN. In addition, in the CNN-PCA formulation presented here, a convolutional neural network is trained as an explicit transform function that can post-process PCA models quickly. CNN-PCA is shown to provide both unconditional and conditional realizations that honor the geological features present in reference SGeMS geostatistical realizations for a binary channelized system. Flow statistics obtained through simulation of random CNN-PCA models closely match results for random SGeMS models for a demanding case in which O-PCA models lead to significant discrepancies. Results for history matching are also presented. In this assessment CNN-PCA is applied with derivative-free optimization, and a subspace randomized maximum likelihood method is used to provide multiple posterior models. Data assimilation and significant uncertainty reduction are achieved for existing wells, and physically reasonable predictions are also obtained for new wells. Finally, the CNN-PCA method is extended to a more complex non-stationary bimodal deltaic fan system, and is shown to provide high-quality realizations for this challenging example.

Yu Sang, Laixi Shi, Yimin Liu

In this paper, we propose a micro hand gesture recognition system and methods using ultrasonic active sensing. This system uses micro dynamic hand gestures for recognition to achieve human-computer interaction (HCI). The implemented system, called hand-ultrasonic gesture (HUG), consists of ultrasonic active sensing, pulsed radar signal processing, and time-sequence pattern recognition by machine learning. We adopt lower frequency (300 kHz) ultrasonic active sensing to obtain high resolution range-Doppler image features. Using high quality sequential range-Doppler features, we propose a state-transition-based hidden Markov model for gesture recognition. This method achieves a recognition accuracy of nearly 90\% by using symbolized range-Doppler features and significantly reduces the computational complexity and power consumption. Furthermore, to achieve higher classification accuracy, we utilize an end-to-end neural network model and obtain a recognition accuracy of 96.32\%. In addition to offline analysis, a real-time prototype is released to verify our method's potential for application in the real world.

Tong Wang, Cynthia Rudin, Finale Doshi-Velez et al.

We present a machine learning algorithm for building classifiers that are comprised of a small number of disjunctions of conjunctions (or's of and's). An example of a classifier of this form is as follows: If X satisfies (x1 = 'blue' AND x3 = 'middle') OR (x1 = 'blue' AND x2 = '<15') OR (x1 = 'yellow'), then we predict that Y=1, ELSE predict Y=0. An attribute-value pair is called a literal and a conjunction of literals is called a pattern. Models of this form have the advantage of being interpretable to human experts, since they produce a set of conditions that concisely describe a specific class. We present two probabilistic models for forming a pattern set, one with a Beta-Binomial prior, and the other with Poisson priors. In both cases, there are prior parameters that the user can set to encourage the model to have a desired size and shape, to conform with a domain-specific definition of interpretability. We provide two scalable MAP inference approaches: a pattern level search, which involves association rule mining, and a literal level search. We show stronger priors reduce computation. We apply the Bayesian Or's of And's (BOA) model to predict user behavior with respect to in-vehicle context-aware personalized recommender systems.