Searcharxiv⌕ Search

arXiv subjects

Xiong Liu

Publications and source records attributed to Xiong Liu.

At least 19 recordsLinked to original sources
arXiv2026

Dynamics of entanglement in non-Hermitian system of nonreciprocal coupling

The exploration of the effect of non-Hermitian (NH) in quantum systems is gaining renewed momentum as a result of the recent progress in experiments. It has been indicated that the ability to engineer the orders of exceptional points (EPs) can be fundamental to its applications in quantum control, such as accelerating entanglement generation; however, an initial survey of the interactions between different orders of EPs is still missing. In this work, we try to partially address this issue by employing a non-reciprocally coupled cavity system which can be experimentally realizable with a two-dimensional reconfigurable NH gauged laser array. With deliberate design, our systems display both second and third orders of EPs. The sign of a novel, purely quantum continuous entanglement phase transition is observed, which reveals how multi-EP interactions alter entanglement dynamics across different regimes while maintaining accelerated generation. Our findings reveil the genuinely quantum NH physics of higher-order EPs, which has opened up a new avenue for future investigations of the higher-order degeneracy application and the dynamic phase transitions of quantum systems.

quant-ph↗
arXiv2026

Plume Segmentation from MethaneSAT with Cross-Sensor Transfer Learning and Physics-Informed Postprocessing

Automated detection and masking of individual methane plumes from satellite imagery is important for operational emission attribution and quantification. We present a machine learning framework for plume detection from MethaneSAT retrieved column-averaged dry-air mole fractions of methane. We address two core challenges: the scarcity of labeled MethaneSAT data and the need for inference reliability across diverse atmospheric and surface conditions. We first demonstrate that Mask R-CNN with a ResNet-50 backbone outperforms U-Net semantic segmentation on both MethaneAIR (an airborne version of MethaneSAT) and MethaneSAT data, with pixel-level F1 score gains of 10.49 and 5.48 respectively. To address MethaneSAT data scarcity, we evaluate three cross-sensor transfer strategies leveraging MethaneAIR flights and synthetic plumes. Mask R-CNN with ResNet-50 fine-tuned from MethaneAIR pre-trained weights is the most effective strategy, achieving instance-level precision of 0.60 and a near-perfect recall of 0.98 at the baseline operating point. A physics-informed post-processing pipeline converts detections into two operationally distinct modes. The first is a high-sensitivity mode that applies morphological filtering and proximity-based merging for comprehensive emission screening, achieving precision of 0.71 and recall of 0.94. The second is a high-precision mode that additionally applies a distribution-based classifier for confident source attribution, achieving precision of 0.92 and recall of 0.70. Manual review of detections classified as false positives against our wavelet-based ground truth labels reveals that a meaningful fraction of cases correspond to real methane enhancements excluded by conservative labeling criteria, indicating that precision values reported are lower bounds on true detection performance... Our data and code are available at: https://doi.org/10.7910/DVN/FR959H

cs.CV↗
arXiv2025

Matrix-Weighted Besov Spaces Associated with Non-isotropic Dilations

Let $\alpha\in\mathbb{R}$, $p\in[1,\infty)$, $q\in(0,\infty]$, $\mathbf{W}$ be a matrix weight, and $A$ be an expansive dilation on $\mathbb{R}^d$. In this paper, the authors firstly investigate and develop some aspects of homogeneous anisotropic Besov spaces $\dot{B}^{\alpha,q}_{p,A}(\mathbb{R}^d,\mathbf{W})$ and inhomogeneous anisotropic Besov spaces $B^{\alpha,q}_{p,A}(\mathbb{R}^d,\mathbf{W})$ theory in the matrix weight setting. Moreover, we show that these spaces are characterized by the magnitude of the $\varphi$-transforms in appropriate sequence spaces. Notably, all these results remain novel even in the diagonal non-isotropic case (when $A = \mathrm{diag}(\lambda_1, \lambda_2, \ldots, \lambda_d)$ with $\{\lambda_j\}_{j=1}^d \subset \mathbb{C}$).

math.FA↗
arXiv2025

Deep Learning for Clouds and Cloud Shadow Segmentation in Methane Satellite and Airborne Imaging Spectroscopy

Effective cloud and cloud shadow detection is a critical prerequisite for accurate retrieval of concentrations of atmospheric methane (CH4) or other trace gases in hyperspectral remote sensing. This challenge is especially pertinent for MethaneSAT, a satellite mission launched in March 2024, to fill a significant data gap in terms of resolution, precision and swath between coarse-resolution global mappers and fine-scale point-source imagers of methane, and for its airborne companion mission, MethaneAIR. MethaneSAT delivers hyperspectral data at an intermediate spatial resolution (approx. 100 x 400, m), whereas MethaneAIR provides even finer resolution (approx. 25 m), enabling the development of highly detailed maps of concentrations that enable quantification of both the sources and rates of emissions. In this study, we use machine learning methods to address the cloud and cloud shadow detection problem for sensors with these high spatial resolutions. Cloud and cloud shadows in remote sensing data need to be effectively screened out as they bias methane retrievals in remote sensing imagery and impact the quantification of emissions. We deploy and evaluate conventional techniques-including Iterative Logistic Regression (ILR) and Multilayer Perceptron (MLP)-with advanced deep learning architectures, namely U-Net and a Spectral Channel Attention Network (SCAN) method. Our results show that conventional methods struggle with spatial coherence and boundary definition, affecting the detection of clouds and cloud shadows. Deep learning models substantially improve detection quality: U-Net performs best in preserving spatial structure, while SCAN excels at capturing fine boundary details... Our data and code is publicly available at: https://doi.org/10.7910/DVN/IKLZOJ

cs.CV↗
arXiv2025

Estimates for Schr\"{o}dinger Groups and Imaginary Power Operators on Weak Hardy Spaces Associated with Non-negative Self-adjoint Operators and Ball Quasi-Banach Function Spaces

Let $(\mathbb{X},d,\mu)$ be a doubling metric measure space, $L$ a non-negative self-adjoint operator on $L^2(\mathbb{X})$ satisfying the Davies-Gaffney estimate, and $X(\mathbb{X})$ a ball quasi-Banach function space on $\mathbb{X}$ satisfying some mild assumptions with $p\in(0,\infty)$ and $s_0\in(0,\min\{p,1\}]$. In this article, the authors study the weak Hardy space $WH_{X,L}(\mathbb{X})$ associated with $L$ and $X(\mathbb{X})$, and then give the atomic and molecular decompositions of $WH_{X,L}(\mathbb{X})$. As applications, the authors establish the boundedness estimate of Schr\"{o}dinger groups for fractional powers of $L$ on $WH_{X,L}(\mathbb{X})$: $$\left\|(I+L)^{-\beta/2}e^{i\tau L^{\gamma/2}}f\right\|_{WH_{X,L}(\mathbb{X})}\leq C\left(1+|\tau|\right)^{n(\frac{1}{s_0}-\frac{r}{2})}\|f\|_{WH_{X,L}(\mathbb{X})},$$ where $0<\gamma\neq1$, $\beta\in[\gamma n(\frac{1}{s_0}-\frac{1}{2}),\infty)$, $r\in(0,1]$, $\tau\in \mathbb{R}$, and $C>0$ is a constant. Moreover, when $(\mathbb{X},d,\mu)$ is an Ahlfors $n$-regular metric measure space and $L$ satisfies the Gaussian upper bound estimate, the authors also obtain the boundedness estimate of imaginary power operators of $L$ on $WH_{X,L}(\mathbb{X})$: $$\left\|L^{i\tau}f\right\|_{WH_{X,L}(\mathbb{X})}\leq C\left(1+|\tau|\right)^{n(\frac{1}{s_0}-\frac{r}{2})}\|f\|_{WH_{X,L}(\mathbb{X})},$$ where $\alpha>n(\frac{1}{s_0}-\frac{1}{2})$, $r\in(\frac{n/s_0}{\alpha+n/2},1]$, $\tau\in \mathbb{R}$, and $C>0$ is a constant. These results are also novelty for strong Hardy spaces $H_{X,L}(\mathbb{X})$. Moreover, all these results have a wide range of generality and, particularly, even when they are applied to weighted Lebesgue spaces, mixed-norm Lebesgue spaces, Orlicz spaces, variable Lebesgue spaces and Euclidean spaces setting, these results are also new.

math.CA↗
arXiv2025

On the Performance of Cloud-based ARM SVE for Zero-Knowledge Proving Systems

Zero-knowledge proofs (ZKP) are becoming a gold standard in scaling blockchains and bringing Web3 to life. At the same time, ZKP for transactions running on the Ethereum Virtual Machine require powerful servers with hundreds of CPU cores. The current zkProver implementation from Polygon is optimized for x86-64 CPUs by vectorizing key operations, such as Merkle tree building with Poseidon hashes over the Goldilocks field, with Advanced Vector Extensions (AVX and AVX512). With these optimizations, a ZKP for a batch of transactions is generated in less than two minutes. With the advent of cloud servers with ARM which are at least 10% cheaper than x86-64 servers and the implementation of ARM Scalable Vector Extension (SVE), we wonder if ARM servers can take over their x86-64 counterparts. Unfortunately, our analysis shows that current ARM CPUs are not a match for their x86-64 competitors. Graviton4 from Amazon Web Services (AWS) and Axion from Google Cloud Platform (GCP) are 1.6X and 1.4X slower compared to the latest AMD EPYC and Intel Xeon servers from AWS with AVX and AVX512, respectively, when building a Merkle tree with over four million leaves. This low performance is due to (1) smaller vector size in these ARM CPUs (128 bits versus 512 bits in AVX512) and (2) lower clock frequency. On the other hand, ARM SVE/SVE2 Instruction Set Architecture (ISA) is at least as powerful as AVX/AVX512 but more flexible. Moreover, we estimate that increasing the vector size to 512 bits will enable higher performance in ARM CPUs compared to their x86-64 counterparts while maintaining their price advantage.

cs.DC↗
arXiv2025

Act-Adaptive Margin: Dynamically Calibrating Reward Models for Subjective Ambiguity

Currently, most reinforcement learning tasks focus on domains like mathematics and programming, where verification is relatively straightforward. However, in subjective tasks such as role-playing, alignment techniques struggle to make progress, primarily because subjective reward modeling using the Bradley-Terry model faces significant challenges when dealing with ambiguous preferences. To improve reward modeling in subjective tasks, this paper proposes AAM (\textbf{\underline{A}}ct-\textbf{\underline{A}}daptive \textbf{\underline{M}}argin), which enhances reward modeling by dynamically calibrating preference margins using the model's internal parameter knowledge. We design two versions of AAM that efficiently generate contextually-appropriate preference gaps without additional human annotation. This approach fundamentally improves how reward models handle subjective rewards by better integrating generative understanding with preference scoring. To validate AAM's effectiveness in subjective reward modeling, we conduct evaluations on RewardBench, JudgeBench, and challenging role-playing tasks. Results show that AAM significantly improves subjective reward modeling performance, enhancing Bradley-Terry reward models by 2.95\% in general tasks and 4.85\% in subjective role-playing tasks. Furthermore, reward models trained with AAM can help downstream alignment tasks achieve better results. Our test results show that applying rewards generated by AAM-Augmented RM to preference learning techniques (e.g., GRPO) achieves state-of-the-art results on CharacterEval and Charm. Code and dataset are available at https://github.com/calubkk/AAM.

cs.CL↗
arXiv2025

OmniCharacter: Towards Immersive Role-Playing Agents with Seamless Speech-Language Personality Interaction

Role-Playing Agents (RPAs), benefiting from large language models, is an emerging interactive AI system that simulates roles or characters with diverse personalities. However, existing methods primarily focus on mimicking dialogues among roles in textual form, neglecting the role's voice traits (e.g., voice style and emotions) as playing a crucial effect in interaction, which tends to be more immersive experiences in realistic scenarios. Towards this goal, we propose OmniCharacter, a first seamless speech-language personality interaction model to achieve immersive RPAs with low latency. Specifically, OmniCharacter enables agents to consistently exhibit role-specific personality traits and vocal traits throughout the interaction, enabling a mixture of speech and language responses. To align the model with speech-language scenarios, we construct a dataset named OmniCharacter-10K, which involves more distinctive characters (20), richly contextualized multi-round dialogue (10K), and dynamic speech response (135K). Experimental results showcase that our method yields better responses in terms of both content and style compared to existing RPAs and mainstream speech-language models, with a response latency as low as 289ms. Code and dataset are available at https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/OmniCharacter.

cs.CL↗
arXiv2025

High-resolution geostationary satellite observations of free tropospheric NO2 over North America: implications for lightning emissions

Free tropospheric (FT) nitrogen dioxide (NO2) plays a critical role in atmospheric oxidant chemistry as a source of tropospheric ozone and of the hydroxyl radical (OH). It also contributes significantly to satellite-observed tropospheric NO2 columns, and must be subtracted when using these columns to quantify surface emissions of nitrogen oxide radicals (NOx = NO + NO2). But large uncertainties remain in the sources and chemistry of FT NO2 because observations are sparse. Here, we construct a new cloud-sliced FT NO2 (700-300 hPa) product from the TEMPO geostationary satellite instrument over North America. This product provides higher data density and quality than previous products from low Earth orbit (LEO) instruments, with the first observation of the FT NO2 diurnal cycle across seasons. Combined with coincident observations from the Geostationary Lightning Mapper (GLM), the TEMPO data demonstrate the dominance of lightning as a source of FT NO2 in non-winter seasons. Comparison of TEMPO FT NO2 data with the GEOS-CF atmospheric chemistry model shows overall consistent magnitudes, seasonality, and diurnal variation, with a midday minimum in non-winter seasons from photochemical loss. However, there are major discrepancies that we attribute to GEOS-CF's use of a standard cloud-top-height (CTH)-based scheme for the lightning NOx source. We find this scheme greatly underestimates offshore lighting flash density and misrepresents the diurnal cycle of lightning over land. Our FT NO2 product provides a unique resource for improving the lightning NOx parameterization in atmospheric models and the ability to use NO2 observations from space to quantify surface NOx emissions.

physics.ao-ph↗
arXiv2025

Local Hardy Spaces Associated with Ball Quasi-Banach Function Spaces and Non-negative Self-adjoint Operators on Spaces of Homogeneous Type and Their Applications

Let $(\mathbb X,d,\mu)$ be a space of homogeneous type in the sense of Coifman--Weiss, let $X$ be a ball quasi-Banach function space on $\mathbb X$ under suitable maximal-function and associate-space assumptions, and let $L$ be a non-negative self-adjoint operator on $L^2(\mathbb X)$. Assume that, for every $t>0$, the semigroup $e^{-tL}$ admits an integral kernel satisfying a Gaussian upper bound. In this paper, we introduce and systematically study the local Hardy space $h_L^X(\mathbb X)$ associated with $X$ and $L$, defined in terms of a local Lusin area function together with an appropriate low-frequency term. As applications of this theory, we establish the boundedness of the local Riesz transform $\nabla(L+\kappa I)^{-1/2}$ from $h_L^X(\mathbb R^d)$ into the corresponding $X$-valued vector function space for second-order divergence-form elliptic operators. We also obtain a H\"{o}rmander-type spectral multiplier theorem for $F(L+I)$ on $h_L^X(\mathbb X)$. Finally, the abstract results are applied to local Orlicz-Hardy spaces, local variable Hardy spaces, and local mixed-norm Hardy spaces. This theory develops Goldberg's original local Hardy space theory [Duke Math. J. {\bf 46} (1979), 27-42; MR0523600] to the setting of ball quasi-Banach function spaces and non-negative self-adjoint operators on spaces of homogeneous type. To the best of our knowledge, several of the results obtained in this paper are new even in the Euclidean setting $\mathbb X:=\mathbb R^d$.

math.FA↗
arXiv2025

OpenOmni: Advancing Open-Source Omnimodal Large Language Models with Progressive Multimodal Alignment and Real-Time Self-Aware Emotional Speech Synthesis

Recent advancements in omnimodal learning have significantly improved understanding and generation across images, text, and speech, yet these developments remain predominantly confined to proprietary models. The lack of high-quality omnimodal datasets and the challenges of real-time emotional speech synthesis have notably hindered progress in open-source research. To address these limitations, we introduce \name, a two-stage training framework that integrates omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model undergoes further training on text-image tasks, enabling (near) zero-shot generalization from vision to speech, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder is trained on speech tasks with direct preference optimization, enabling real-time emotional speech synthesis with high fidelity. Experiments show that \name surpasses state-of-the-art models across omnimodal, vision-language, and speech-language benchmarks. It achieves a 4-point absolute improvement on OmniBench over the leading open-source model VITA, despite using 5x fewer training samples and a smaller model size (7B vs. 7x8B). Additionally, \name achieves real-time speech generation with <1s latency at non-autoregressive mode, reducing inference time by 5x compared to autoregressive methods, and improves emotion classification accuracy by 7.7\%

cs.CL↗
arXiv2024

Real-variable Theory of Anisotropic Musielak-Orlicz-Lorentz Hardy Spaces with Applications to Calder\'{o}n-Zygmund Operators

Let $\varphi: \mathbb{R}^{n}\times[0,\infty)\rightarrow[0,\infty)$ be a Musielak-Orlicz function satisfying the uniformly anisotropic Muckenhoupt condition and be of uniformly lower type $p^-_{\varphi}$ and of uniformly upper type $p^+_{\varphi}$ with $0<p^-_{\varphi}\leq p^+_{\varphi}<\infty$, $q\in(0,\infty]$, and $A$ be a general expansive matrix on $\mathbb{R}^{n}$. In this article, the authors first introduce the anisotropic Musielak-Orlicz-Lorentz Hardy space $H^{\varphi,q}_A(\mathbb{R}^{n})$ which, when $q=\infty$, coincides with the known anisotropic weak Musielak-Orlicz Hardy space $H^{\varphi,\infty}_A(\mathbb{R}^{n})$, and then establish atomic and molecular characterizations of $H^{\varphi,q}_A(\mathbb{R}^{n})$. As applications, the authors prove the boundedness of anisotropic Calder\'{o}n-Zygmund operators on $H^{\varphi,q}_A(\mathbb{R}^{n})$ when $q\in(0,\infty)$ or from the anisotropic Musielak-Orlicz Hardy space $H^{\varphi}_A(\mathbb{R}^{n})$ to $H^{\varphi,\infty}_A(\mathbb{R}^{n})$ in the critical case. The ranges of all the exponents under consideration are the best possible admissible ones which particularly improve all the known corresponding results for $H^{\varphi,\infty}_A(\mathbb{R}^{n})$ via widening the original assumption $0<p^-_{\varphi}\leq p^+_{\varphi}\leq1$ into the full range $0<p^-_{\varphi}\leq p^+_{\varphi}<\infty$, and all the results when $q\in(0,\infty)$ are new and generalized from isotropic setting to anisotropic setting.

math.CA↗
arXiv2024

MMEvol: Empowering Multimodal Large Language Models with Evol-Instruct

The development of Multimodal Large Language Models (MLLMs) has seen significant advancements with increasing demands in various fields (e.g., multimodal agents, embodied intelligence). While model-driven approaches attempt to enhance MLLMs capabilities through diverse architectures, the gains have become increasingly marginal. Conversely, data-driven methods, which scale up image-text instruction data, are more effective but face limited data diversity and complexity challenges. The absence of high-quality data constitutes a significant development barrier for MLLMs. To address the data quality bottleneck, we propose MMEvol, a novel multimodal instruction data evolution framework. This framework iteratively improve data quality through a refined combination of fine-grained perception, cognitive reasoning, and interaction evolution, generating a more complex and diverse image-text instruction dataset that empowers MLLMs with enhanced capabilities. Beginning with an initial set of instructions, SEED-163K, we utilize MMEvol to systematically broaden the diversity of instruction types, extend visual reasoning steps to improve cognitive reasoning abilities, and thoroughly explore fine-grained information within images to enhance visual understanding and robustness. To comprehensively evaluate the effectiveness of our approach, we conduct extensive qualitative analysis and quantitative experiments across 13 vision-language tasks. Compared to baseline models trained with the initial seed data, the results demonstrate that our method achieves an average accuracy improvement of 3.1 percentage points. Furthermore, our approach reaches state-of-the-art (SOTA) performance in nine tasks using significantly less data compared to state-of-the-art models.

cs.CL↗
arXiv2024

Ozone Anomalies in Dry Intrusions Associated with Atmospheric Rivers

As a result of their important role in weather and the global hydrological cycle, understanding atmospheric rivers' (ARs) connection to synoptic-scale climate patterns and atmospheric dynamics has become increasingly important. In addition to case studies of two extreme AR events, we produce a December climatology of the three-dimensional structure of water vapor and O3 (ozone) distributions associated with ARs in the northeastern Pacific from 2004-2014 using MERRA-2 reanalysis products. Results show that positive O3 anomalies reside in dry intrusions of stratospheric air due to stratosphere-to-troposphere transport (STT) behind the intense water vapor transport of the AR. In composites, we find increased excesses of O3 concentration, as well as in the total O3 flux within the dry intrusions, with increased AR strength. We find that STT O3 flux associated with ARs over the NE Pacific accounts for up to 13 percent of total Northern Hemisphere STT O3 flux in December, and extrapolation indicates that AR-associated dry intrusions may account for as much as 32 percent of total NH STT O3 flux. This study quantifies STT of O3 in connection with ARs for the first time and improves estimates of tropospheric ozone concentration due to STT in the identification of this correlation. In light of predictions that ARs will become more intense and/or frequent with climate change, quantifying AR-related STT O3 flux is especially valuable for future radiative forcing calculations.

physics.ao-ph↗
arXiv2023

Improving Factual Consistency of News Summarization by Contrastive Preference Optimization

Despite the recent progress in news summarization made by large language models (LLMs), they often generate summaries that are factually inconsistent with original articles, known as "hallucinations" in text generation. Unlike previous small models (e.g., BART, T5), current LLMs make fewer silly mistakes but more sophisticated ones, such as imposing cause and effect, adding false details, overgeneralizing, etc. These hallucinations are challenging to detect through traditional methods, which poses great challenges for improving the factual consistency of text summarization. In this paper, we propose Contrastive Preference Optimization (CPO) to disentangle the LLMs' propensities to generate faithful and fake content. Furthermore, we adopt a probing-based specific training method to improve their capacity of distinguishing two types of propensities. In this way, LLMs can execute the instructions more accurately and have enhanced perception of hallucinations. Experimental results show that CPO significantly improves the reliability of summarization based on LLMs.

cs.CL↗
arXiv2023

Clinical Trial Recommendations Using Semantics-Based Inductive Inference and Knowledge Graph Embeddings

Designing a new clinical trial entails many decisions, such as defining a cohort and setting the study objectives to name a few, and therefore can benefit from recommendations based on exhaustive mining of past clinical trial records. Here, we propose a novel recommendation methodology, based on neural embeddings trained on a first-of-a-kind knowledge graph of clinical trials. We addressed several important research questions in this context, including designing a knowledge graph (KG) for clinical trial data, effectiveness of various KG embedding (KGE) methods for it, a novel inductive inference using KGE, and its use in generating recommendations for clinical trial design. We used publicly available data from clinicaltrials.gov for the study. Results show that our recommendations approach achieves relevance scores of 70%-83%, measured as the text similarity to actual clinical trial elements, and the most relevant recommendation can be found near the top of list. Our study also suggests potential improvement in training KGE using node semantics.

cs.AI↗
arXiv2023

Meta-lenses for differential imaging based on weak measurement

All-optical information communication, processing and computation have received substantial interest of both fundamental and applied research due to its unrivaled speed and broad bandwidth. Compared to its electronic counterpart, photons seldom interact with each other which makes them obtain a long coherence time on one hand and relieved from heavy energy dissipation on the other. However, one of the hindrances to achieve all-optical circuits is the large volume of all-optical devices to achieve specific functionalities. In this work, we propose and demonstrate experimentally three meta-lenses for differential imaging employing the framework of weak measurement: (1) partial differential lens, (2) total differential lens and (3) second order differential lens compatible with the requirement of miniaturization to achieve all-optical technology. Based on Fresnel-lens-like structures, our meta-lenses incorporated the previous weak-measurement compartment into wavelength scale, which induces a miniature differential operation system as a result. In addition to its potential importance in heavily integrated all-optical neural networks, the differential lens can be easily incorporated in the existing imaging systems like a conventional lens without increasing the complexity of the system of interest.

physics.optics↗
arXiv2022

Customizing Knowledge Graph Embedding to Improve Clinical Study Recommendation

Inferring knowledge from clinical trials using knowledge graph embedding is an emerging area. However, customizing graph embeddings for different use cases remains a significant challenge. We propose custom2vec, an algorithmic framework to customize graph embeddings by incorporating user preferences in training the embeddings. It captures user preferences by adding custom nodes and links derived from manually vetted results of a separate information retrieval method. We propose a joint learning objective to preserve the original network structure while incorporating the user's custom annotations. We hypothesize that the custom training improves user-expected predictions, for example, in link prediction tasks. We demonstrate the effectiveness of custom2vec for clinical trials related to non-small cell lung cancer (NSCLC) with two customization scenarios: recommending immuno-oncology trials evaluating PD-1 inhibitors and exploring similar trials that compare new therapies with a standard of care. The results show that custom2vec training achieves better performance than the conventional training methods. Our approach is a novel way to customize knowledge graph embeddings and enable more accurate recommendations and predictions.

cs.LG↗