An information-rich sampling technique over spatio-temporal CNN for classification of human actions in videos.

We propose a novel video sampling scheme for human action recognition in videos, using Gaussian Weighing Function. Traditionally in deep learning-based human activity recognition approaches, either a few random frames or every k t h frame of the video is considered for training the 3D CNN, where k is a small positive integer, like 4, 5, or 6. This kind of sampling reduces the volume of the input data, which speeds-up the training network and also avoids overfitting to some extent, thus enhancing the performance of the 3D CNN model. In the proposed video sampling technique, consecutive k frames of a video are aggregated into a single frame by computing a Gaussian-weighted summation of the k frames. The resulting frame preserves the information in a better way than the conventional approaches and experimentally shown to perform better. In this paper, a 3-Dimensional deep CNN is proposed to extract the spatio-temporal features and follows Long Short-Term Memory (LSTM) to recognize human actions. The proposed 3D CNN architecture is capable of handling the videos where the camera is placed at a distance from the performer. Experiments are performed with KTH, WEIZMANN, and CASIA-B Human Activity and Gait datasets, whereby it is shown to outperform state-of-the-art deep learning based techniques. We achieve 95.78%, 95.27%, and 95.27% over the KTH, WEIZMANN, and CASIA-B human action and gait recognition datasets, respectively.

HRel: Filter pruning based on High Relevance between activation maps and class labels.

This paper proposes an Information Bottleneck theory based filter pruning method that uses a statistical measure called Mutual Information (MI). The MI between filters and class labels, also called Relevance, is computed using the filter's activation maps and the annotations. The filters having High Relevance (HRel) are considered to be more important. Consequently, the least important filters, which have lower Mutual Information with the class labels, are pruned. Unlike the existing MI based pruning methods, the proposed method determines the significance of the filters purely based on their corresponding activation map's relationship with the class labels. Architectures such as LeNet-5, VGG-16, ResNet-56, ResNet-110 and ResNet-50 are utilized to demonstrate the efficacy of the proposed pruning method over MNIST, CIFAR-10 and ImageNet datasets. The proposed method shows the state-of-the-art pruning results for LeNet-5, VGG-16, ResNet-56, ResNet-110 and ResNet-50 architectures. In the experiments, we prune 97.98%, 84.85%, 76.89%, 76.95%, and 63.99% of Floating Point Operation (FLOP)s from LeNet-5, VGG-16, ResNet-56, ResNet-110, and ResNet-50 respectively. The proposed HRel pruning method outperforms recent state-of-the-art filter pruning methods. Even after pruning the filters from convolutional layers of LeNet-5 drastically (i.e., from 20, 50 to 2, 3, respectively), only a small accuracy drop of 0.52% is observed. Notably, for VGG-16, 94.98% parameters are reduced, only with a drop of 0.36% in top-1 accuracy. ResNet-50 has shown a 1.17% drop in the top-5 accuracy after pruning 66.42% of the FLOPs. In addition to pruning, the Information Plane dynamics of Information Bottleneck theory is analyzed for various Convolutional Neural Network architectures with the effect of pruning. The code is available at https://github.com/sarvanichinthapalli/HRel.

SVEP1 is important for morphogenesis of lymphatic system: Possible implications in lymphedema.

SVEP1, also known as Polydom, is a large extracellular mosaic protein with functions in protein interactions and adhesion. Since Svep1 knockout animals show severe edema and lymphatic system malformations, the aim of this study is to evaluate the presence of SVEP1 variants in patients with lymphedema. We analyzed DNA from 246 lymphedema patients for variants in known lymphedema genes, 235 of whom tested negative and underwent a second testing for new candidate genes, including SVEP1, as reported here. We found three samples with rare heterozygous missense single-nucleotide variants in the SVEP1 gene. In one family, healthy members were found to carry the same variants and reported some subclinical edema. Based on our findings and a review of the literature, we propose SVEP1 as a candidate gene that should be sequenced in patients with lymphatic malformations, with or without lymphedema, in order to investigate and add evidence on its possible involvement in the development of lymphedema.

AutoTune: Automatically Tuning Convolutional Neural Networks for Improved Transfer Learning.

Transfer learning enables solving a specific task having limited data by using the pre-trained deep networks trained on large-scale datasets. Typically, while transferring the learned knowledge from source task to the target task, the last few layers are fine-tuned (re-trained) over the target dataset. However, these layers are originally designed for the source task that might not be suitable for the target task. In this paper, we introduce a mechanism for automatically tuning the Convolutional Neural Networks (CNN) for improved transfer learning. The pre-trained CNN layers are tuned with the knowledge from target data using Bayesian Optimization. First, we train the final layer of the base CNN model by replacing the number of neurons in the softmax layer with the number of classes involved in the target task. Next, the CNN is tuned automatically by observing the classification performance on the validation data (greedy criteria). To evaluate the performance of the proposed method, experiments are conducted on three benchmark datasets, e.g., CalTech-101, CalTech-256, and Stanford Dogs. The classification results obtained through the proposed AutoTune method outperforms the standard baseline transfer learning methods over the three datasets by achieving 95.92%, 86.54%, and 84.67% accuracy over CalTech-101, CalTech-256, and Stanford Dogs, respectively. The experimental results obtained in this study depict that tuning of the pre-trained CNN layers with the knowledge from the target dataset confesses better transfer learning ability. The source codes are available at https://github.com/JekyllAndHyde8999/AutoTune_CNN_TransferLearning.

Rare PECAM1 variants in three families with lymphedema.

PECAM1 is a member of the immunoglobulin superfamily and is expressed in monocytes, neutrophils, macrophages and other types of immune cells as well as in endothelial cells. PECAM1 function is crucial for the development and maturation of B lymphocytes. The aim of this study was to link rare PECAM1 variants found in lymphedema patients with the development of lymphatic system malformations. Using NGS, we previously tested 246 Italian lymphedema patients for variants in 29 lymphedema-associated genes and obtained 235 negative results. We then tested these patients for variants in the PECAM1 gene. We found three probands with rare variants in PECAM1. All variants were heterozygous missense variants. In Family 1, the unaffected mother and brother of the proband were found to carry the same variant as the proband. Lymphoscintigraphy was performed to determine possible lymphatic malformations and showed that in both cases a bilateral slight reduction in the speed and lymphatic clearance of the lower limbs. PECAM1 function is important for lymphatic vasculature formation. We found variants in PECAM1 that may be associated with susceptibility to lymphedema.

Review of the function of SEMA3A in lymphatic vessel maturation and its potential as a candidate gene for lymphedema: Analysis of three families with rare causative variants.

SEMA3A is a semaphorin involved in cell signaling with PlexinA1 and Neuropilin-1 (NRP1) receptors and it is responsible for recruiting dendritic cells into lymphatics. Mutations in the SEMA3A gene result in abnormalities in lymphatic vessel development and maturation. We investigated the association of SEMA3A variants detected in lymphedema patients with lymphatic maturation and lymphatic system malfunction. First, we used NGS technology to sequence the SEMA3A gene in 235 lymphedema patients who carry wild type alleles for known lymphedema genes. We detected three different missense variants in three families. Bioinformatic results showed that some protein interactions could be altered by these variants. Other unaffected family members of the probands also reported different episodes of subclinical edema. We then evaluated the importance of the SEMA3A gene in the formation and maturation of lymphatic vessels. Our results determined that SEMA3A variants segregate in families with lymphatic system malformations and recommend the inclusion of SEMA3A in the gene panel for testing of patients with lymphedema.

CYP26B1 and its implications in lymphangiogenesis: Literature review and study of rare variants in two families.

CYP26B1 is a member of the cytochrome P450 family and is responsible for the break-down of retinoic acid for which appropriate levels are important for normal development of the cardiovascular and lymphatic systems. In a cohort of 235 patients with lymphatic malformations, we performed genetic testing for the CYP26B1 gene. These probands had previously tested negative for known lymphedema genes. We identified two heterozygous missense CY-P26B1 variants in two patients. Our bioinformatic study suggested that alterations caused by these variants have no major effect on the overall stability of CYP26B1 protein structure. Balanced levels of retinoic acid maintained by CYP26B1 are crucial for the lymphatic system. We identified that CYP26B1 could be involved in predisposition for lymphedema. We propose that CYP26B1 be further explored as a new candidate gene for genetic testing of lymphedema patients.