FedTKD: A Trustworthy Heterogeneous Federated Learning Based on Adaptive Knowledge Distillation.

Federated learning allows multiple parties to train models while jointly protecting user privacy. However, traditional federated learning requires each client to have the same model structure to fuse the global model. In real-world scenarios, each client may need to develop personalized models based on its environment, making it difficult to perform federated learning in a heterogeneous model environment. Some knowledge distillation methods address the problem of heterogeneous model fusion to some extent. However, these methods assume that each client is trustworthy. Some clients may produce malicious or low-quality knowledge, making it difficult to aggregate trustworthy knowledge in a heterogeneous environment. To address these challenges, we propose a trustworthy heterogeneous federated learning framework (FedTKD) to achieve client identification and trustworthy knowledge fusion. Firstly, we propose a malicious client identification method based on client logit features, which can exclude malicious information in fusing global logit. Then, we propose a selectivity knowledge fusion method to achieve high-quality global logit computation. Additionally, we propose an adaptive knowledge distillation method to improve the accuracy of knowledge transfer from the server side to the client side. Finally, we design different attack and data distribution scenarios to validate our method. The experiment shows that our method outperforms the baseline methods, showing stable performance in all attack scenarios and achieving an accuracy improvement of 2% to 3% in different data distributions.

Emotion Recognition from Chinese Speech for Smart Affective Services Using a Combination of SVM and DBN.

Accurate emotion recognition from speech is important for applications like smart health care, smart entertainment, and other smart services. High accuracy emotion recognition from Chinese speech is challenging due to the complexities of the Chinese language. In this paper, we explore how to improve the accuracy of speech emotion recognition, including speech signal feature extraction and emotion classification methods. Five types of features are extracted from a speech sample: mel frequency cepstrum coefficient (MFCC), pitch, formant, short-term zero-crossing rate and short-term energy. By comparing statistical features with deep features extracted by a Deep Belief Network (DBN), we attempt to find the best features to identify the emotion status for speech. We propose a novel classification method that combines DBN and SVM (support vector machine) instead of using only one of them. In addition, a conjugate gradient method is applied to train DBN in order to speed up the training process. Gender-dependent experiments are conducted using an emotional speech database created by the Chinese Academy of Sciences. The results show that DBN features can reflect emotion status better than artificial features, and our new classification approach achieves an accuracy of 95.8%, which is higher than using either DBN or SVM separately. Results also show that DBN can work very well for small training databases if it is properly designed.

Mechanism of improving effect of losartan on insulin sensitivity of non-insulin-dependent diabetes mellitus rats.

The specific inhibition of angiotensin II action at AT(1) receptors by losartan has been shown to decrease peripheral insulin resistance in type 2 diabetic patients and animal models. We examined the effect of losartan on the expression of insulin receptor substrate 1 (IRS-1), protein kinase B (PKB) and glucose transporter 4 (GLUT4), as well as the phosphorylation status of IRS-1 and the association between IRS-1 and phosphatidylinositol (PI) 3-kinase in skeletal muscle from fat-fed and-streptozotocin (STZ)-treated rats, an animal model of type 2 diabetes mellitus. In addition, the effects of losartan on GLUT4 translocation in muscle cells and on insulin sensitivity were also evaluated. Muscle tissues were isolated from male losartan-treated and untreated normal or non-insulin-dependent diabetes mellitus (NIDDM) rats with a dose of 4 mg/kg per day for 6 weeks. Oral administration of losartan improved insulin sensitivity, which was determined by an oral glucose tolerance test (OGTT). In skeletal muscles, the protein levels of IRS-1, PKB and GLUT4 in NIDDM rats were not significantly different from those of the control rats, and they were not affected by losartan. The levels of IRS-1 tyrosine phosphorylation, PI 3-kinase activity associated with IRS-1 and PKB activation after stimulation with insulin in muscle tissue of NIDDM rats were significantly decreased (P<0.01) compared with those in the control rats, while they were not increased by losartan. Losartan had a major effect on GLUT4 translocation in myocytes, as it significantly increased (P<0.05) the insulin-induced amounts of GLUT4 in plasma membrane (PM) and T-tubules (TT) in myocytes from NIDDM rats. Consistent with these results, the plasma glucose level in losartan-treated NIDDM rats was decreased (P<0.05) compared with that in untreated NIDDM rats. Our results suggest that losartan may exert beneficial effects on insulin resistance by increasing the translocation of GLUT4 in muscle tissue, which is probably associated with a non-PI 3-kinase-dependent mechanism.