arXiv subjects

Li Jingjing

Publications and source records attributed to Li Jingjing.

5 recordsLinked to original sources

arXiv2026

OnomaCompass: A Texture Exploration Interface that Shuttles between Words and Images

Humans can finely perceive material textures, yet articulating such somatic impressions in words is a cognitive bottleneck in design ideation. We present OnomaCompass, a web-based exploration system that links sound-symbolic onomatopoeia and visual texture representations to support early-stage material discovery. Instead of requiring users to craft precise prompts for generative AI, OnomaCompass provides two coordinated latent-space maps--one for texture images and one for onomatopoeic term--built from an authored dataset of invented onomatopoeia and corresponding textures generated via Stable Diffusion. Users can navigate both spaces, trigger cross-modal highlighting, curate findings in a gallery, and preview textures applied to objects via an image-editing model. The system also supports video interpolation between selected textures and re-embedding of extracted frames to form an emergent exploration loop. We conducted a within-subjects study with 11 participants comparing OnomaCompass to a prompt-based image-generation workflow using Gemini 2.5 Flash Image ("Nano Banana"). OnomaCompass significantly reduced workload (NASA-TLX overall, mental demand, effort, and frustration; p < .05) and increased hedonic user experience (UEQ), while usability (SUS) favored the baseline. Qualitative findings indicate that OnomaCompass helps users externalize vague sensory expectations and promotes serendipitous discovery, but also reveals interaction challenges in spatial navigation. Overall, leveraging sound symbolism as a lightweight cue offers a complementary approach to Kansei-driven material ideation beyond prompt-centric generation.

cs.HC↗

arXiv2022

Neutral Stellar Winds Toward the High-Mass Star-Forming Region G176.51+00.20

We observed the high-mass star-forming region G176.51+00.20 using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) with the 19-beam tracking observational mode. This is a pilot work of searching for neutral stellar winds traced by atomic hydrogen (i.e., HI winds) using the high sensitivity HI line toward high-mass star-forming regions where bipolar molecular outflows have been detected with high sensitivity by Liu et al. HI wind was detected in this work only in Beam 1. We find here that, similar to low-mass star formation, no matter how large the inclination is, the HI wind is likely sufficiently strong to drive a molecular outflow. We also find that the abundance of HI in the HI wind is consistent with that of the HI narrow-line self-absorption (HINSA) in the same beam (i.e., Beam 1). This implies that there is probably an internal relationship between HI winds and HINSA. This result also reinforces the assertion that HI winds and detected molecular outflows are associated with each other.

astro-ph.GA↗

arXiv2022

HI Narrow-Line Self-Absorptions Toward the High-Mass Star-Forming Region G176.51+00.20

Using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) 19-beam tracking observational mode, high sensitivity and high-velocity resolution HI spectral lines have been observed toward the high-mass star-forming region G176.51+00.20. This is a pilot study of searching for HI narrow-line self-absorption (HINSA) toward high-mass star-forming regions where bipolar molecular outflows have been detected. This work is confined to the central seven beams of FAST. Two HINSA components are detected in all seven beams, which correspond to a strong CO emission region (SCER; with a velocity of $\sim$ $-$18 km s$^{-1}$) and a weak CO emission region (WCER; with a velocity of $\sim$ $-$3 km s$^{-1}$). The SCER detected in Beam 3 is probably more suitably classified as a WCER. In the SCER, the HINSA is probably associated with the molecular material traced by the CO. The fractional abundance of HINSA ranges from $\sim 1.1 \times 10^{-3}$ to $\sim 2.6 \times 10^{-2}$. Moreover, the abundance of HINSA in Beam 1 is lower than that in the surrounding beams (i.e., Beams 2 and 4--7). This possible ring could be caused by ionization of HI or relatively rapid conversion from HI to H$_2$ in the higher-density inner region. In the WCER (including Beam 3 in the SCER), the HINSA is probably not associated with CO clouds, but with CO-dark or CO-faint gas.

astro-ph.GA↗

arXiv2020

Maximum Density Divergence for Domain Adaptation

Unsupervised domain adaptation addresses the problem of transferring knowledge from a well-labeled source domain to an unlabeled target domain where the two domains have distinctive data distributions. Thus, the essence of domain adaptation is to mitigate the distribution divergence between the two domains. The state-of-the-art methods practice this very idea by either conducting adversarial training or minimizing a metric which defines the distribution gaps. In this paper, we propose a new domain adaptation method named Adversarial Tight Match (ATM) which enjoys the benefits of both adversarial training and metric learning. Specifically, at first, we propose a novel distance loss, named Maximum Density Divergence (MDD), to quantify the distribution divergence. MDD minimizes the inter-domain divergence ("match" in ATM) and maximizes the intra-class density ("tight" in ATM). Then, to address the equilibrium challenge issue in adversarial domain adaptation, we consider leveraging the proposed MDD into adversarial domain adaptation framework. At last, we tailor the proposed MDD as a practical learning loss and report our ATM. Both empirical evaluation and theoretical analysis are reported to verify the effectiveness of the proposed method. The experimental results on four benchmarks, both classical and large-scale, show that our method is able to achieve new state-of-the-art performance on most evaluations. Codes and datasets used in this paper are available at {\it github.com/lijin118/ATM}.

cs.CV↗

arXiv2019

Locality Preserving Joint Transfer for Domain Adaptation

Domain adaptation aims to leverage knowledge from a well-labeled source domain to a poorly-labeled target domain. A majority of existing works transfer the knowledge at either feature level or sample level. Recent researches reveal that both of the paradigms are essentially important, and optimizing one of them can reinforce the other. Inspired by this, we propose a novel approach to jointly exploit feature adaptation with distribution matching and sample adaptation with landmark selection. During the knowledge transfer, we also take the local consistency between samples into consideration, so that the manifold structures of samples can be preserved. At last, we deploy label propagation to predict the categories of new instances. Notably, our approach is suitable for both homogeneous and heterogeneous domain adaptation by learning domain-specific projections. Extensive experiments on five open benchmarks, which consist of both standard and large-scale datasets, verify that our approach can significantly outperform not only conventional approaches but also end-to-end deep models. The experiments also demonstrate that we can leverage handcrafted features to promote the accuracy on deep features by heterogeneous adaptation.

cs.CV↗