arXiv subjects

Haonan Deng

Publications and source records attributed to Haonan Deng.

2 recordsLinked to original sources

arXiv2026

Do SpeechLMs Hear Their Own Opinions? Diagnosing and Mitigating Previous-Belief Contamination in Streaming Emotion Understanding

Streaming emotion understanding uses historical state while continuously interpreting current audio, often feeding the model's previous prediction back as context. We show that this history conditioning can distort current perception. On a balanced CREMA-D-Stream counterfactual diagnostic, changing only the injected previous emotion label while holding the audio fixed reduces current-audio accuracy from 72.50% to 30.42% and flips 65.69% of predictions. The effect is strongly label-asymmetric, with prior pull ranging from 4.76% to 98.20%, revealing a failure we call previous-belief contamination (PBC). To address PBC, we introduce EmoUpdate, a training-free framework that separates current-audio perception from historical state revision through three components: (1) a prior-blind acoustic firewall that prevents historical state from entering perception; (2) an evidence-shrunk causal belief filter that introduces history only after observation formation and retains label-asymmetric transition structure only when supported by observed evidence; and (3) a closed-form decontamination operator derived from the same counterfactual measurements for serving stacks where firewalling is unavailable. Across four SpeechLMs and two streaming emotion benchmarks, EmoUpdate achieves the best step accuracy and state-balanced accuracy in all eight model--benchmark settings, improving S-BAcc by up to 69.71 points and step accuracy by up to 38.41 points over the strongest controlled baselines.

cs.SD↗

arXiv2018

An Efficient Privacy-Preserving Algorithm based on Randomized Response in IoT-based Smart Grid

Among existing privacy-preserving approaches, Differential Privacy (DP) is a powerful tool that can provide privacy-preserving noisy query answers over statistical databases and has been widely adopted in many practical fields. In particular, as a privacy machine of DP, Randomized Aggregable Privacy-Preserving Ordinal Response (RAPPOR) enables strong privacy, efficient, and high-utility guarantees for each client string in data crowdsourcing. However, as for Internet of Things(IoT), such as smart gird, data are often processed in batches. Therefore, developing a new random response algorithm that can support batch-processing tend to make it more efficient and suitable for IoT applications than existing random response algorithms. In this paper, we propose a new randomized response algorithm that can achieve differential-privacy and utility guar-antees for consumer's behaviors, and process a batch of data at each time. Firstly, by applying sparse coding in this algorithm, a behavior signature dictionary is created from the aggregated energy consumption data in fog. Then, we add noise into the behavior signature dictionary by classical randomized response techniques and achieve the differential privacy after data re-aggregation. Through the security analysis with the principle of differential privacy and experimental results verification, we find that our Algorithm can preserve consumer's privacy with-out comprising utility.

cs.CR↗