Searcharxiv⌕ Search

arXiv subjects

Boxuan Shan

Publications and source records attributed to Boxuan Shan.

2 recordsLinked to original sources

EdgeNet City: Context-Aware AI-RAN Planning toward Autonomous Urban Connectivity

The evolution from 5G toward 6G calls for more anticipatory and context-aware Cell-on-Wheels (COW) planning in wireless networks. We present EdgeNet City, an online framework that grounds free-form natural-language inputs in geographic activity representations and path-gain data. A catalog-referenced representation links activity roles, geographic features and predefined demand profiles, separating semantic selection from deterministic spatial computation. The algorithm combines prescribed demand with relative propagation support to screen service areas for subsequent COW assessment. We publicly release a London dataset with 22,678 transmitters, approximately 449.4 billion path-gain records and retained receiver geometry. Within prepared propagation domains, the online prototype reuses these records without new ray tracing or live operator measurements. In a matched 20-event study across six London environments, the expanded representation increases mean spatial overlap with the reused full technical-review reference from 0.4929 for the single-feature Venue-only baseline to 0.7203. Rule-based and language-model planners select identical activities in this panel. With need-aware ranking and a three-cell diagnostic budget fixed, the expanded representation increases mean Reference-need Capture by 25.44 percentage points over Venue-only.

cs.NI↗

ReAGent: A Model-agnostic Feature Attribution Method for Generative Language Models

Feature attribution methods (FAs), such as gradients and attention, are widely employed approaches to derive the importance of all input features to the model predictions. Existing work in natural language processing has mostly focused on developing and testing FAs for encoder-only language models (LMs) in classification tasks. However, it is unknown if it is faithful to use these FAs for decoder-only models on text generation, due to the inherent differences between model architectures and task settings respectively. Moreover, previous work has demonstrated that there is no `one-wins-all' FA across models and tasks. This makes the selection of a FA computationally expensive for large LMs since input importance derivation often requires multiple forward and backward passes including gradient computations that might be prohibitive even with access to large compute. To address these issues, we present a model-agnostic FA for generative LMs called Recursive Attribution Generator (ReAGent). Our method updates the token importance distribution in a recursive manner. For each update, we compute the difference in the probability distribution over the vocabulary for predicting the next token between using the original input and using a modified version where a part of the input is replaced with RoBERTa predictions. Our intuition is that replacing an important token in the context should have resulted in a larger change in the model's confidence in predicting the token than replacing an unimportant token. Our method can be universally applied to any generative LM without accessing internal model weights or additional training and fine-tuning, as most other FAs require. We extensively compare the faithfulness of ReAGent with seven popular FAs across six decoder-only LMs of various sizes. The results show that our method consistently provides more faithful token importance distributions.

cs.CL↗