yokoxue.github.io/research.html at master · yokoxue/yokoxue.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
  "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<meta name="generator" content="jemdoc, see http://jemdoc.jaboc.net/" />
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
<link rel="stylesheet" href="jemdoc.css" type="text/css" />
<title>RESEARCH AREAS</title>
</head>
<body>
<table summary="Table for page layout." id="tlayout">
<tr valign="top">
<td id="layout-menu">
<div class="menu-category">Menu</div>
<div class="menu-item"><a href="home.html">Home</a></div>
<div class="menu-item"><a href="aboutme.html">About&nbsp;Me</a></div>
<div class="menu-item"><a href="research.html" class="current">Research</a></div>
<div class="menu-item"><a href="publications.html">Publications</a></div>
<div class="menu-item"><a href="teaching.html">Teaching</a></div>
<div class="menu-item"><a href="Group.html">Members and photos</a></div>
<div class="menu-category">Others</div>
<div class="menu-item"><a href="resources.html">Resources</a></div>
</td>
<td id="layout-content">
<div id="toptitle">
<h1>RESEARCH AREAS</h1>
</div>
<div class="infoblock">
<div class="blockcontent">
<p>I work on   Efficient and Explainable artificial intelligence (XAI), World Models (WM) for electromagnetics/wireless, and Learning-based Optimization (L2O) for real-world communication and network decisions. Our goal is interpretable, editable, and deployable methods that connect physics, data, and decision making.</p>
</div></div>
<h2>Theoretical Part</h2>
<h3>Sparse and Efficient AI</h3>
<p>Structured sparsity, low-rank, and tensorization across data, features, models, and gradients, with unified compression and distillation for edge–cloud collaboration and federated training. HW/SW co-acceleration (CPU, GPU, FPGA) enables reliable on-device deployment in wireless applications.</p>
<p>Structured sparsity and low-rank modeling</p>

<ul>
<li>
<ul>
<li><p>Compression and distillation for models and gradients
</p>
</li>
<li><p>Communication-efficient training and adaptation
</p>
</li>
<li><p>Algorithm–system co-design for throughput and energy gains</p>
</li>
</ul>

</li>
</ul>


<h3>Explainable AI (XAI)</h3>
<p>Principled interpretability and reliability for models: from attribution and concept-based explanations to causal and statistical guarantees, ensuring trustworthy deployment.</p>

<p>Interpretable modeling and guarantees</p>
<ul>
  <li>
    <ul>
      <li><p>Attribution and concept-based explanations; counterfactual and causal probes; manifold learning
</p></li>
      <li><p>Generalization and convergence bounds for nonconvex/structured models
</p></li>
      <li><p>Risk control with conformal prediction; safety constraints and robust training</p></li>
    </ul>
  </li>
</ul>
<h2>Application Part</h2>
<h3>World Models (WM) for Radio Environment</h3>
<p>Physically grounded, data-driven models of radio EM environments that fuse geometry and materials (e.g., BIM, point clouds, maps), contextual signals (vision, language, IMU, GNSS), and RF measurements (sweeps arrays). We combine neural representations (3DGS and NeRF variants, diffusion and energy models, GNNs, neural operators such as FNO, DeepONet, and GNO) with PDE and boundary consistency, constitutive relations, and structured sparsity or low-rank priors. The models support editable inversion, uncertainty quantification, and active re-measurement for localized statistical channel modeling and environment-aware communication.</p>
<p>Localized statistical channel modeling</p>

<ul>
<li>
<ul>
<li><p>Neural radio radiance fields; structured Bayesian inference on graphs
/p>
</li>
<li><p>PDE and boundary consistency; material priors; calibrated uncertainty
with multimodal fusion</p>
</li>
<li><p>Geometry/materials plus RF arrays plus contextual signals</p>
</li>
</ul>

</li>
</ul>
<h3>Learning-based Optimization (L2O)</h3>
<p>Differentiable, learning-augmented solvers (RL, GNN, and neural combinatorial optimization) operating on WM for constrained decisions, with feasibility guarantees, learned warm-starts, and fast cross-scenario adaptation.</p>
<p>Communication and RAN optimization</p>

<ul>
<li>
<ul>
<li><p>UAV sensing plan, site placement; beam, power, and spectrum allocation; routing and formation planning
</p>
</li>
<li><p>Relax–optimize–and–sample; neural projections; constraint satisfaction
deployment</p>
</li>
<li><p>Edge and federated execution for real-time, certifiable decisions</p>
</li>
</ul>

</li>
</ul>
<h3>FieldMind: From Representation to Decisions</h3>
<p>FieldMind is our instantiation of WM plus sparse and efficient AI plus L2O, coupled with algorithm–hardware co-design (C<tt></tt>, CUDA, FPGA; edge and federated) for online decision making. It maps structure across data, algorithms, and architectures into solvers and hardware. </p>
</td>
</tr>
</table>
</body>
</html>