AMRD：用於輕量級語音情感辨識的自適應多教師關係蒸餾

摘要

裝置端語音情緒辨識（SER）對即時應用至關重要，然而在 SER 上表現優異的大型自監督模型對邊緣裝置而言成本過高。多教師知識蒸餾可將這些模型壓縮為輕量級學生模型，但仍存在兩項挑戰：教師模型的可靠性會因批次而異，且 logit 層級的蒸餾忽略了樣本間的關係結構。我們提出自適應多教師關係蒸餾（AMRD）來處理這兩個問題。透過對每個教師的 logit 相似度矩陣採用單類別支援向量機，指派每批次權重，傾向於一致性較高的教師。關係蒸餾損失函數對齊教師與學生的相似度矩陣，捕捉 logit 匹配所遺漏的結構。在 IEMOCAP 與 CREMA-D 資料集上，跨越四種學生架構，AMRD 在多數設定下優於單一教師蒸餾的基線方法；消融實驗亦證實兩項組成能產生互補的增益。

English

On-device speech emotion recognition (SER) is critical for real-time applications, yet large self-supervised models that excel at SER are too costly for edge devices. Multi-teacher knowledge distillation can compress them into a lightweight student, but two challenges remain: teacher reliability varies across batches, and logit-level distillation ignores inter-sample relational structure. We propose Adaptive Multi-teacher Relational Distillation (AMRD) to address both. A one-class SVM on each teacher's logit similarity matrix assigns per-batch weights favoring more coherent teachers. A relational distillation loss aligns teacher and student similarity matrices, capturing structure that logit matching misses. On IEMOCAP and CREMA-D datasets across four student architectures, AMRD outperforms single-teacher distillation baselines in most settings, and ablations confirm both components yield complementary gains.