Efficient Gradient Updating Strategies with Adaptive Power Allocation for Federated Learning over Wireless Backhaul.

In this paper, efficient gradient updating strategies are developed for the federated learning when distributed clients are connected to the server via a wireless backhaul link. Specifically, a common convolutional neural network (CNN) module is shared for all the distributed clients and it is trained through the federated learning over wireless backhaul connected to the main server. However, during the training phase, local gradients need to be transferred from multiple clients to the server over wireless backhaul link and can be distorted due to wireless channel fading. To overcome it, an efficient gradient updating method is proposed, in which the gradients are combined such that the effective SNR is maximized at the server. In addition, when the backhaul links for all clients have small channel gain simultaneously, the server may have severely distorted gradient vectors. Accordingly, we also propose a binary gradient updating strategy based on thresholding in which the round associated with all channels having small channel gains is excluded from federated learning. Because each client has limited transmission power, it is effective to allocate more power on the channel slots carrying specific important information, rather than allocating power equally to all channel resources (equivalently, slots). Accordingly, we also propose an adaptive power allocation method, in which each client allocates its transmit power proportionally to the magnitude of the gradient information. This is because, when training a deep learning model, the gradient elements with large values imply the large change of weight to decrease the loss function.

Linear Discriminant Analysis-Based Motion Classification Using Distributed Micro-Doppler Radars with Limited Backhaul.

In this paper, we propose a cooperative linear discriminant analysis (LDA)-based motion classification algorithm for distributed micro-Doppler (MD) radars which are connected to a data fusion center through the limited backhaul. Due to the limited backhaul, each radar cannot report the high-dimensional data of a multi-aspect angle MD signature to the fusion center. Instead, at each radar, the dimensionality of the MD signature is reduced by using the LDA algorithm and the dimensionally-reduced MD signature can be collected at the data fusion center. To further reduce the burden of backhaul, we also propose the softmax processing method in which the distances of the sensed MD signatures from the centers of clusters for all motion candidates are computed at each radar. The output of the softmax process at each radar is quantized through the pyramid vector quantization with a finite number of bits and is reported to the data fusion center. To improve the classification performance at the fusion center, the channel resources of the backhaul are adaptively allocated based on the classification separability at each radar. The proposed classification performance was assessed with synthetic simulation data as well as experimental data measured through the USRP-based MD radar.

Genetic G2548A polymorphism of leptin gene and risk of cancer: a meta-analysis of 6860 cases and 7956 controls.

PURPOSE: The results from the published studies on the association between LEP (leptin) genetic polymorphism and cancer risk are conflicting. The common G2548A genetic polymorphism has been reported to be functional and may contribute to genetic susceptibility to cancers. However, the association between LEP G2548A genetic polymorphism and cancer risk remains inconclusive. METHODS: To better understand the role of LEP G2548A genetic polymorphism in global cancer, we conducted this comprehensive meta-analysis encompassing 6860 cases and 7956 controls. RESULTS: Overall, the LEP G2548A genetic polymorphism was associated with higher cancer risk in three genetic models (AA vs GG, AA vs AG/GG, A vs G). In the stratified analysis, there was significant association of LEP G2548A variant with non-Hodgkin's lymphoma (NHL) under homozygous co-dominant model (OR=1.27, 95% CI 1.01-1.60) and additive genetic model (OR=1.14, 95% CI 1.01-1.28). Moreover, a significantly increased cancer risk was found in three genetic models (AA vs GG, AA vs AG/GG, A vs G) among Caucasian population. For Asians, no significant associations were observed in any genetic model tested. CONCLUSIONS: These findings suggest that the LEP G2548A genetic polymorphism may increase the susceptibility of cancers in NHL, especially in the homozygote co-dominant model and the additive genetic model among Caucasian populations. The phenomenon also indicates that the SNP functions as a recessive mutation, which needs to be verified or linked with functional studies.

A study on the mechanical characteristics of the EBM-printed Ti-6Al-4V LCP plates in vitro.

PURPOSE: The electron beam melting (EBM) Ti-6Al-4V material technology has been developed over a short time period. It was introduced through a research to develop Ti-6Al-4V implants for patients, but EBM printed locking compression plates have not been used for clinical implants. The main purpose of this study is to find whether the EBM Ti-6Al-4V plate suit for clinical implants. METHODS: First, we scanned an AO-locking compression plate (LCP) and printed LCP samples using EBM. Next, we evaluated the EBM plate surface roughness through optical microscopy as well as the LCP and EBM plates' mechanical characteristics using the ASTM standard, which is commonly used to test the mechanical properties of bone plates subject to bending. Each sample was examined using a single-cycle four-point bending test and hardness testing to acquire data on bending stiffness, bending strength, bending structural stiffness, and hardness. RESULTS: The results show significant differences in bending stiffness, bending strength, bending structural stiffness, and hardness between the samples using EBM and the original LCP plates. The EBM-printed samples' surface roughness was 0.49 ± 0.02 µm. The mean hardness of the LCP sample was 266.67 HV10 ± 5.8, and the EBM-printed sample mean hardness was 341.1 HV10 ± 1.93. The EBM samples' bending stiffness was 87.67%, which is greater than using the LCP plates'; and the bending strength was 190.7% greater, the bending structural stiffness was 73.2% greater, and the hardness was 27.9% greater. CONCLUSIONS: The results show that the EBM plates' general mechanical strength was significantly greater than the LCP plates. An EBM plate is advantageous for clinical implants because it can be customized with great potential for improvement.

Lack of association between LEPR Lys656Asn or Ser343Ser polymorphisms and cancer susceptibility: A meta-analysis.

Previous studies have shown conflicting results between the association of leptin receptor (LEPR) genetic polymorphisms and cancer risk. The frequent LEPR Lys656Asn or Ser343Ser genetic polymorphism has been demonstrated to be functional and may promote genetic susceptibility to cancers. However, the association between the LEPR Lys656Asn or Ser343Ser genetic polymorphism and cancer risk remains to be determined. To improve the understanding of the LEPR Lys656Asn or Ser343Ser genetic polymorphism role in global cancer, a comprehensive meta-analysis was conducted that comprised 2,480 cases and 3,162 controls. The LEPR Lys656Asn or Ser343Ser genetic polymorphism did not significantly affect the cancer risk. In the stratified analysis, there was no significant association of the LEPR Lys656Asn or Ser343Ser variants with any type of cancer under any model. In addition, significantly increased risks were found in the Asian population in heterozygous codominant [odds ratio (OR), 1.24 (1.01-1.53)] and dominant [OR, 1.24 (1.02-1.50)] genetic models. A significantly increased susceptibility to cancer was not found when stratified by study design. There were no significant differences found in genotype method and sample size in cases among the genotypes. These findings indicated a lack of association between LEPR Lys656Asn or Ser343Ser polymorphisms and cancer susceptibility, however, these polymorphisms may increase the cancer susceptibility among the Asian population, particularly in the dominant genetic model. The single-nucleotide polymorphism is also suggested to function as a dominant mutation, which requires verification or association with functional studies.

Association of LEPR K109R polymorphisms with cancer risk: a systematic review and pooled analysis.

PURPOSE: This meta-analysis was conducted to evaluate the association between LEPR K109R (rs1137100) genetic polymorphism and cancer risk. METHODS: To better understand the role of LEPR K109R(rs1137100) genetic polymorphism in global cancer, we conducted this comprehensive meta-analysis encompassing 5819 cases and 8068 controls. RESULTS: Overall, the LEPR K109R(rs1137100) genetic polymorphism did not significantly affect the risk of cancer. In the stratified analysis, significant associations were found between the LEPR K109R(rs1137100) genetic polymorphism and breast cancer under additive genetic model (odds ratio/OR=0.67, 95% CI 0.61-0.73). For prostate cancer, there was no significant association of LEPR K109R(rs1137100) variant with this disease under any model. For lung cancer, there was significant association of LEPR K109R(rs1137100) variant with the disease under heterozygous co-dominant model (OR=0.72, 95% CI 0.55- 0.96), recessive genetic model (OR=0.76, 95% CI 0.61-0.94) and additive genetic model (OR=0.89, 95% CI 0.80-0.99). For gastric cancer, significant association was found in the 3 genetic models (AG vs GG, AA/AG vs GG and A vs G), the ORS (95%CI) being 2.93 (1.25-6.86), 2.93 (1.25-6.86) and 2.25 (1.07-4.72), respectively. Moreover, no significant cancer risk was found in any genetic model among Caucasian and Asian populations. When stratified by study design, no significantly elevated susceptibility to cancer was found among any studies. No significant differences in the genotype method and sample size in cases were found among genotypes. CONCLUSION: These findings suggested that the LEPR K109R(rs1137100) genetic polymorphism may decrease the susceptibility in breast cancer, especially in the additive genetic model. The findings also indicate that single nucleotide polymorphism (SNP) functions as a recessive mutation, which needs to be verified or linked with functional studies.

Lack of association between LEPR Q223R polymorphisms and cancer susceptibility: evidence from a meta-analysis.

PURPOSE: The results from the published studies on the association between LEPR genetic polymorphisms and cancer risk are conflicting. The common LEPR Q223R genetic polymorphism has been reported to be functional and may contribute to genetic susceptibility to cancer. However, the association between LEPR Q223R genetic polymorphism and cancer risk remains inconclusive. METHODS: To better understand the role of LEPR Q223R genetic polymorphism in global cancer, we conducted this comprehensive meta-analysis encompassing 9139 cases and 11282 controls. RESULTS: Overall, the LEPR Q223R genetic polymorphism did not significantly affect the risk of cancer. In the stratified analysis, there was no significant association of LEPR Q223R variant with breast cancer, colorectal cancer and non-Hodgkin's lymphoma (NHL) under any models. Moreover, significantly increased risks were found in Asian and African in all genetic models tested. When stratified by study design, no significantly increased susceptibility to cancer was found among any studies. No significantly differences in sample size in cases were found among genotypes. CONCLUSIONS: These findings suggested lack of association between LEPR Q223R polymorphisms and cancer susceptibility, but the LEPR Q223R genetic polymorphism may increase the susceptibility to cancers in Asian and African individuals. Large, well designed epidemiological studies are needed to validate our findings.

Association of LEP A19G polymorphism with cancer risk: a systematic review and pooled analysis.

The results from the published studies on the association between leptin (LEP) genetic polymorphism and cancer risk are conflicting. The common A19G (rs2167270) genetic polymorphism has been reported to be functional and may contribute to genetic susceptibility to cancers. However, the association between LEP A19G (rs2167270) genetic polymorphism and cancer risk remains inconclusive. To better understand the role of LEP A19G (rs2167270) genetic polymorphism in global cancer, we conducted this comprehensive meta-analysis encompassing 5,679 cases and 7,443 controls. Overall, the LEP A19G (rs2167270) genetic polymorphism was associated with lower cancer risk. In the stratified analysis, significant associations were found between the LEP A19G (rs2167270) genetic polymorphism and colorectal cancer and non-Hodgkin's lymphoma. For colorectal cancer, there was no significant association of LEP A19G (rs2167270) variant with this disease under heterozygous codominant model [odds ratio (OR) = 1.11 (0.97-1.27)], dominant genetic model [OR = 1.03 (0.91-1.17)], and additive genetic model [OR = 0.94 (0.86-1.03)]; however, there was a marginal association under homozygous codominant model [OR = 0.80 (0.66-0.97)] and recessive genetic model [OR = 0.75 (0.63-0.90)]. For non-Hodgkin's lymphoma, there was a significant association of LEP A19G (rs2167270) variant with the disease under homozygous codominant model [OR = 0.74 (0.59-0.94)], recessive genetic model [OR = 0.76 (0.61-0.94)], and additive genetic model [OR = 0.89 (0.80-0.99)], but not under heterozygous codominant model [OR = 0.95 (0.82-1.10)] and dominant genetic model [OR = 0.91 (0.79-1.04)]. Moreover, a significantly decreased cancer risk was found in recessive genetic model among Latin American population. When stratified by study design, significantly elevated susceptibility to cancer was not found among any studies. No significantly differences in genotype method and sample size in cases were found among genotypes. These findings suggest that the LEP A19G (rs2167270) genetic polymorphism may decrease the susceptibility to cancers in colorectal cancer and non-Hodgkin's lymphoma, when assuming a homozygote codominant model and a recessive genetic model among Latin American population. The phenomenon also indicates that the SNP functions as a recessive mutation, which needs to be verified or linked with functional studies.

Maohuoside A acts in a BMP-dependent manner during osteogenesis.

There has been a strong interest in searching for natural therapies for osteoporosis. Epimedium koreanum Nakai is an herb that is commonly used in East Asia to treat osteoporosis, and most studies of its activity have focused on its major ingredient, icariin. In this study, maohuoside A (MHA), a single compound isolated from the E. koreanum, was found to promote osteogenesis in mouse bone marrow-derived mesenchymal stem cells. We hypothesise that, if MHA potently induces osteogenic differentiation in a bone morphogenetic protein-dependent manner, it may be used to broaden the sources for cell transplantation and thereby establish more efficient bone regeneration systems. Real-time polymerase chain reaction and western blot were used to detect the expression of SMAD4, a marker of bone formation. Microcomputed tomography and histomorphometric techniques showed that oral treatment with MHA was followed by an increase in the bone mineral density of the lumbar vertebrae in mice. This result also indicates that MHA may directly activate osteopontin gene transcription. In conclusion, MHA seems to enhance the osteogenesis of bone marrow-derived mesenchymal stem cells at least partly via bone morphogenetic protein signalling.