arXiv subjects

Kaili Wang

Publications and source records attributed to Kaili Wang.

At least 19 recordsLinked to original sources

arXiv2026

KernelArc: A Multi-Agent Framework for GPU Kernel Optimization

We present KernelArc, a multi-agent framework for autonomous GPU kernel optimization across heterogeneous workloads. Strategy-specialized agents run in parallel and coordinate through conclusions-only shared memory, a deterministic benchmark guard, and read-only cross-agent state with plateau-triggered drafting. We evaluate KernelArc on NVIDIA H100 and B200 GPUs using category-representative SOL-ExecBench workloads. The resulting implementations span custom BF16 GEMM, static cuBLASLt Expert-API configuration tables, fused mixture-of-experts backward, shape-gated decoder-layer fusion, native NVFP4 grouped-query attention, and paged prefill attention. In the public SOL-ExecBench leaderboard snapshot recorded on August~20, 2026, KernelArc ranked first on every representative L1, L2, Quantization, and FlashInfer task evaluated. The trajectories support the paper's central motivation: shared multi-agent search can broaden exploration and reach stronger incumbents within a fixed candidate budget, while the value of individual coordination features depends on the kernel and optimization stage.

cs.AI↗

arXiv2026

Addressing Detail Bottlenecks in Latent Diffusion for RGB-to-SWIR Image Translation

Latent diffusion models (LDMs) enable efficient image-to-image translation but discard fine spatial details during compression, degrading downstream perception tasks. We identify two bottlenecks: the autoencoder, which loses spatial information, and the conditioning pathway, which further degrades the source signal through naive downsampling. We propose two lightweight, backbone-agnostic fixes: a Source-Conditioned Autoencoder (SCAE) that injects high-resolution source features into the decoder via skip connections, and a Learnable Guidance Encoder (LGE) that replaces naive downsampling with a learned conditioning signal. Evaluated on RGB-to-SWIR translation for driving scenes with two denoiser backbones (U-Net and DiT), our approach improves detection mAP by up to 2x over the latent diffusion baseline, with up to 3.4x gains on small objects (COCO-small, <32^2 px^2), while achieving state-of-the-art FID. We further show that FID and detection performance are poorly correlated, motivating multi-axis evaluation. Results generalise zero-shot to the public RASMD benchmark. We will publicly release test data with annotations, all checkpoints, and training code.

cs.CV↗

arXiv2025

Increasing the Diversity in RGB-to-Thermal Image Translation for Automotive Applications

Thermal imaging in Advanced Driver Assistance Systems (ADAS) improves road safety with superior perception in low-light and harsh weather conditions compared to traditional RGB cameras. However, research in this area faces challenges due to limited dataset availability and poor representation in driving simulators. RGB-to-thermal image translation offers a potential solution, but existing methods focus on one-to-one mappings. We propose a one-to-many mapping using a multi-modal translation framework enhanced with our Component-aware Adaptive Instance Normalization (CoAdaIN). Unlike the original AdaIN, which applies styles globally, CoAdaIN adapts styles to different image components individually. The result, as we show, is more realistic and diverse thermal image translations. This is the accepted author manuscript of the paper published in IEEE Sensors Conference 2024. The final published version is available at 10.1109/SENSORS60989.2024.10785056.

cs.CV↗

arXiv2025

Giant Splitting of Folded Dirac Bands in Kekul\'{e}-ordered Graphene with Eu Intercalation

Kekul\'{e}-ordered graphene on SiC realized by intercalating two-dimensional metal layers offers a versatile platform for exploring intriguing quantum states and phenomena. Here, we achieve the intercalation of $(\mathrm{\sqrt{3}\times\sqrt{3}})\mathit{R}30^\circ$-ordered Eu layer between epitaxial graphene and SiC substrate, realizing a Kekul\'{e} graphene with large local magnetic moments of intercalated Eu atoms. Combining angle-resolved photoemission spectroscopy (ARPES) and density functional theory (DFT) calculations, we revealed that the Kekul{\'{e}} order folds the Dirac cones of graphene from the corners to the Brillouin zone center via intervalley scattering, forming the replica Dirac bands with gap opening. More intriguingly, the Dirac fermions in the replica Dirac bands show a strong exchange coupling with the localized magnetic moments of Eu $4f$ orbitals, resulting in a giant splitting of the folded Dirac bands. The observation of strong coupling between Dirac fermions and local magnetic moments of Eu $4f$ electrons via Kekul\'{e} order pave a new way for generating Dirac band splitting in graphene, advancing the potential applications of Kekul\'{e}-ordered graphene in spintronics, as well as exploring intriguing physical properties and correlation states for quantum technology.

cond-mat.str-el↗

arXiv2024

Robust Coulomb Gap and Varied-temperature Study of Epitaxial 1T'-WSe$_2$ Monolayers

The transition metal dichalcogenides (TMDCs) with a 1T' structural phase are predicted to be two-dimensional topological insulators at zero temperature. Although the quantized edge conductance of 1T'-WTe$_2$ has been confirmed to survive up to 100 K, this temperature is still relatively low for industrial applications. Addressing the limited studies on temperature effects in 1T'-TMDCs, our research focuses on the electronic and crystal properties of the epitaxial 1T'-WSe$_2$ monolayers grown on bilayer graphene (BLG) and SrTiO$_3$(100) substrates at various temperatures. For the 1T'-WSe$_2$ grown on BLG, we observed a significant thermal expansion effect on its band structures with a thermal expansion coefficient of $\sim$60$\times$10$^{-6}$ K$^{-1}$. In contrast, the 1T'-WSe$_2$ grown on SrTiO$_3$(100) exhibits minimal changes with varied temperatures due to the enhanced strain exerted by the substrate. Besides, A significant Coulomb gap (CG) was observed pinned at the Fermi level in the angle-resolved photoemission spectroscopy (ARPES) and scanning tunneling spectroscopy (STS). The CG was founded to decrease with increasing temperatures, and can persist up to 200 K for 1T'-WSe$_2$/BLG, consistent with our Monte Carlo simulations. The robustness of the CG and the positive fundamental gap endow the epitaxial 1T'-WSe$_2$ monolayers with huge potential for realizing the quantum spin Hall devices.

cond-mat.mes-hall↗

arXiv2023

DTIAM: A unified framework for predicting drug-target interactions, binding affinities and activation/inhibition mechanisms

Accurate and robust prediction of drug-target interactions (DTIs) plays a vital role in drug discovery. Despite extensive efforts have been invested in predicting novel DTIs, existing approaches still suffer from insufficient labeled data and cold start problems. More importantly, there is currently a lack of studies focusing on elucidating the mechanism of action (MoA) between drugs and targets. Distinguishing the activation and inhibition mechanisms is critical and challenging in drug development. Here, we introduce a unified framework called DTIAM, which aims to predict interactions, binding affinities, and activation/inhibition mechanisms between drugs and targets. DTIAM learns drug and target representations from large amounts of label-free data through self-supervised pre-training, which accurately extracts the substructure and contextual information of drugs and targets, and thus benefits the downstream prediction based on these representations. DTIAM achieves substantial performance improvement over other state-of-the-art methods in all tasks, particularly in the cold start scenario. Moreover, independent validation demonstrates the strong generalization ability of DTIAM. All these results suggested that DTIAM can provide a practically useful tool for predicting novel DTIs and further distinguishing the MoA of candidate drugs. DTIAM, for the first time, provides a unified framework for accurate and robust prediction of drug-target interactions, binding affinities, and activation/inhibition mechanisms.

q-bio.BM↗

arXiv2023

R-Pool and Settlement Markets for Recoverable ERC-20R Tokens

ERC-20R is a wrapper around ERC-20 that supports asset recovery within a limited time window after an asset is transferred. It is designed to reduce theft and losses on the blockchain by allowing a victim to recover their stolen or lost assets during the recovery window. When an honest recipient receives an ERC-20R asset, they must wait until the recovery windows elapses (say, 24 hours), before they can unwrap the asset back to its base ERC-20 form. We argue that many DeFi services will likely refuse to accept unsettled recoverable assets because they can interfere with their normal operations. Consequently, when Alice receives an ERC-20R token, she must wait 24 hours before she can use it with a DeFi service. But what if Alice is willing to pay a fee to exchange the wrapped token for an unwrapped ERC-20 token that can be used right away? In this paper we explore how to design a pool to exchange an unsettled ERC-20R asset for a base ERC-20 of the same asset. Designing such a pool raises several challenging questions and we present our solutions.

cs.CR↗

arXiv2023

Deriving Weeklong Activity-Travel Dairy from Google Location History: Survey Tool Development and A Field Test in Toronto

This paper introduces an innovative travel survey methodology that utilizes Google Location History (GLH) data to generate travel diaries for transportation demand analysis. By leveraging the accuracy and omnipresence among smartphone users of GLH, the proposed methodology avoids the need for proprietary GPS tracking applications to collect smartphone-based GPS data. This research enhanced an existing travel survey designer, Travel Activity Internet Survey Interface (TRAISI), to make it capable of deriving travel diaries from the respondents' GLH. The feasibility of this data collection approach is showcased through the Google Timeline Travel Survey (GTTS) conducted in the Greater Toronto Area, Canada. The resultant dataset from the GTTS is demographically representative and offers detailed and accurate travel behavioural insights.

econ.GN↗

arXiv2022

ERC-20R and ERC-721R: Reversible Transactions on Ethereum

Blockchains are meant to be persistent: posted transactions are immutable and cannot be changed. When a theft takes place, there are limited options for reversing the disputed transaction, and this has led to significant losses in the blockchain ecosystem. In this paper we propose reversible versions of ERC-20 and ERC-721, the most widely used token standards. With these new standards, a transaction is eligible for reversal for a short period of time after it has been posted on chain. After the dispute period has elapsed, the transaction can no longer be reversed. Within the short dispute period, a sender can request to reverse a transaction by convincing a decentralized set of judges to first freeze the disputed assets, and then later convincing them to reverse the transaction. Supporting reversibility in the context of ERC-20 and ERC-721 raises many interesting technical challenges. This paper explores these challenges and proposes a design for our ERC-20R and ERC-721R standards, the reversible versions of ERC-20 and ERC-721. We also provide a prototype implementation. Our goal is to initiate a deeper conversation about reversibility in the hope of reducing some of the losses in the blockchain ecosystem.

cs.CR↗

arXiv2022

Modelling the Frequency of Home Deliveries: An Induced Travel Demand Contribution of Aggrandized E-shopping in Toronto during COVID-19 Pandemics

The COVID-19 pandemic dramatically catalyzed the proliferation of e-shopping. The dramatic growth of e-shopping will undoubtedly cause significant impacts on travel demand. As a result, transportation modeller's ability to model e-shopping demand is becoming increasingly important. This study developed models to predict household' weekly home delivery frequencies. We used both classical econometric and machine learning techniques to obtain the best model. It is found that socioeconomic factors such as having an online grocery membership, household members' average age, the percentage of male household members, the number of workers in the household and various land use factors influence home delivery demand. This study also compared the interpretations and performances of the machine learning models and the classical econometric model. Agreement is found in the variable's effects identified through the machine learning and econometric models. However, with similar recall accuracy, the ordered probit model, a classical econometric model, can accurately predict the aggregate distribution of household delivery demand. In contrast, both machine learning models failed to match the observed distribution.

econ.EM↗

arXiv2022

Information Compensation for Deep Conditional Generative Networks

In recent years, unsupervised/weakly-supervised conditional generative adversarial networks (GANs) have achieved many successes on the task of modeling and generating data. However, one of their weaknesses lies in their poor ability to separate, or disentangle, the different factors that characterize the representation encoded in their latent space. To address this issue, we propose a novel structure for unsupervised conditional GANs powered by a novel Information Compensation Connection (IC-Connection). The proposed IC-Connection enables GANs to compensate for information loss incurred during deconvolution operations. In addition, to quantify the degree of disentanglement on both discrete and continuous latent variables, we design a novel evaluation procedure. Our empirical results suggest that our method achieves better disentanglement compared to the state-of-the-art GANs in a conditional generation setting.

cs.LG↗

arXiv2022

Tutela: An Open-Source Tool for Assessing User-Privacy on Ethereum and Tornado Cash

A common misconception among blockchain users is that pseudonymity guarantees privacy. The reality is almost the opposite. Every transaction one makes is recorded on a public ledger and reveals information about one's identity. Mixers, such as Tornado Cash, were developed to preserve privacy through "mixing" transactions with those of others in an anonymity pool, making it harder to link deposits and withdrawals from the pool. Unfortunately, it is still possible to reveal information about those in the anonymity pool if users are not careful. We introduce Tutela, an application built on expert heuristics to report the true anonymity of an Ethereum address. In particular, Tutela has three functionalities: first, it clusters together Ethereum addresses based on interaction history such that for an Ethereum address, we can identify other addresses likely owned by the same entity; second, it shows Ethereum users their potentially compromised transactions; third, Tutela computes the true size of the anonymity pool of each Tornado Cash mixer by excluding potentially compromised transactions. A public implementation of Tutela can be found at https://github.com/TutelaLabs/tutela-app. To use Tutela, visit https://www.tutela.xyz.

cs.CR↗

arXiv2021

Audio-Visual Grounding Referring Expression for Robotic Manipulation

Referring expressions are commonly used when referring to a specific target in people's daily dialogue. In this paper, we develop a novel task of audio-visual grounding referring expression for robotic manipulation. The robot leverages both the audio and visual information to understand the referring expression in the given manipulation instruction and the corresponding manipulations are implemented. To solve the proposed task, an audio-visual framework is proposed for visual localization and sound recognition. We have also established a dataset which contains visual data, auditory data and manipulation instructions for evaluation. Finally, extensive experiments are conducted both offline and online to verify the effectiveness of the proposed audio-visual framework. And it is demonstrated that the robot performs better with the audio-visual data than with only the visual data.

cs.RO↗

arXiv2021

An Unpaired Shape Transforming Method for Image Translation and Cross-Domain Retrieval

We address the problem of unpaired geometric image-to-image translation. Rather than transferring the style of an image as a whole, our goal is to translate the geometry of an object as depicted in different domains while preserving its appearance characteristics. Our model is trained in an unpaired fashion, i.e. without the need of paired images during training. It performs all steps of the shape transfer within a single model and without additional post-processing stages. Extensive experiments on the VITON, CMU-Multi-PIE and our own FashionStyle datasets show the effectiveness of the method. In addition, we show that despite their low-dimensionality, the features learned by our model are useful to the item retrieval task.

cs.CV↗

arXiv2021

MinMaxCAM: Improving object coverage for CAM-basedWeakly Supervised Object Localization

One of the most common problems of weakly supervised object localization is that of inaccurate object coverage. In the context of state-of-the-art methods based on Class Activation Mapping, this is caused either by localization maps which focus, exclusively, on the most discriminative region of the objects of interest or by activations occurring in background regions. To address these two problems, we propose two representation regularization mechanisms: Full Region Regularizationwhich tries to maximize the coverage of the localization map inside the object region, and Common Region Regularization which minimizes the activations occurring in background regions. We evaluate the two regularizations on the ImageNet, CUB-200-2011 and OpenImages-segmentation datasets, and show that the proposed regularizations tackle both problems, outperforming the state-of-the-art by a significant margin.

cs.CV↗

arXiv2021

Towards Human-Understandable Visual Explanations:Imperceptible High-frequency Cues Can Better Be Removed

Explainable AI (XAI) methods focus on explaining what a neural network has learned - in other words, identifying the features that are the most influential to the prediction. In this paper, we call them "distinguishing features". However, whether a human can make sense of the generated explanation also depends on the perceptibility of these features to humans. To make sure an explanation is human-understandable, we argue that the capabilities of humans, constrained by the Human Visual System (HVS) and psychophysics, need to be taken into account. We propose the {\em human perceptibility principle for XAI}, stating that, to generate human-understandable explanations, neural networks should be steered towards focusing on human-understandable cues during training. We conduct a case study regarding the classification of real vs. fake face images, where many of the distinguishing features picked up by standard neural networks turn out not to be perceptible to humans. By applying the proposed principle, a neural network with human-understandable explanations is trained which, in a user study, is shown to better align with human intuition. This is likely to make the AI more trustworthy and opens the door to humans learning from machines. In the case study, we specifically investigate and analyze the behaviour of the human-imperceptible high spatial frequency features in neural networks and XAI methods.

cs.CV↗

arXiv2021

In Defense of LSTMs for Addressing Multiple Instance Learning Problems

LSTMs have a proven track record in analyzing sequential data. But what about unordered instance bags, as found under a Multiple Instance Learning (MIL) setting? While not often used for this, we show LSTMs excell under this setting too. In addition, we show thatLSTMs are capable of indirectly capturing instance-level information us-ing only bag-level annotations. Thus, they can be used to learn instance-level models in a weakly supervised manner. Our empirical evaluation on both simplified (MNIST) and realistic (Lookbook and Histopathology) datasets shows that LSTMs are competitive with or even surpass state-of-the-art methods specially designed for handling specific MIL problems. Moreover, we show that their performance on instance-level prediction is close to that of fully-supervised methods.

cs.CV↗

arXiv2021

Multiple Exemplars-based Hallucinationfor Face Super-resolution and Editing

Given a really low-resolution input image of a face (say 16x16 or 8x8 pixels), the goal of this paper is to reconstruct a high-resolution version thereof. This, by itself, is an ill-posed problem, as the high-frequency information is missing in the low-resolution input and needs to be hallucinated, based on prior knowledge about the image content. Rather than relying on a generic face prior, in this paper, we explore the use of a set of exemplars, i.e. other high-resolution images of the same person. These guide the neural network as we condition the output on them. Multiple exemplars work better than a single one. To combine the information from multiple exemplars effectively, we introduce a pixel-wise weight generation module. Besides standard face super-resolution, our method allows to perform subtle face editing simply by replacing the exemplars with another set with different facial features. A user study is conducted and shows the super-resolved images can hardly be distinguished from real images on the CelebA dataset. A qualitative comparison indicates our model outperforms methods proposed in the literature on the CelebA and WebFace dataset.

cs.CV↗