Article
Computer Science, Artificial Intelligence
Peter K. Koo, Matt Ploenzke
Summary: The study shows that using exponential activations in deep convolutional neural networks leads to more interpretable and robust representations of genomic sequence motifs. Better test performance in CNNs does not necessarily mean more interpretable representations. Exponential activations significantly improve the efficacy of recovering biologically meaningful representations.
NATURE MACHINE INTELLIGENCE
(2021)
Article
Biochemical Research Methods
Kerr Ding, Sheng Wang, Yunan Luo
Summary: This article introduces a new approach to biological network alignment called GraNA, which utilizes deep learning and functional correspondence between proteins across species to predict functional relatedness. GraNA performs well on multiple tasks and successfully discovers functionally replaceable human-yeast protein pairs documented in previous studies.
Article
Computer Science, Artificial Intelligence
Jialin Liu, Fei Chao, Chih-Min Lin, Changle Zhou, Changjing Shang
Summary: This paper introduces dynamic kernel convolutional neural networks (DK-CNNs) and explains how they enhance the expressive capacity of convolutional operations by extending a latent dimension. DK convolution analyzes fixed features with a latent variable, leading to better performance compared to regular CNNs.
Article
Computer Science, Artificial Intelligence
Xinxin Shan, Tai Ma, Yutao Shen, Jiafeng Li, Ying Wen
Summary: The study introduces a novel attention convolution method called Kernel Attention Convolution (KAConv) to enhance the flexibility of convolution. By embedding attention into the convolution kernel, KAConv generates different attention weights to dynamically adjust the parameters of convolution kernels, improving the flexibility of convolution. Experiment results demonstrate that KAConv outperforms existing attention mechanism-based methods in the ImageNet-1K benchmark.
Article
Computer Science, Information Systems
Kwanghyun Koo, Hyun Kim
Summary: In this study, a new vectorized structured kernel pruning method is proposed, which achieves high FLOPs reduction and minimal accuracy degradation while maintaining the weight structure. Experimental results demonstrate significant parameter and FLOPs reduction, as well as real acceleration effects on GPUs, in various networks including ResNet-50.
Article
Multidisciplinary Sciences
Feng Pan, Bingyao Huang, Chunhong Zhang, Xinning Zhu, Zhenyu Wu, Moyu Zhang, Yang Ji, Zhanfei Ma, Zhengchen Li
Summary: In this study, a deep learning model called SAVSNet is proposed for student dropout prediction. By smoothing time series and integrating missingness patterns, the model is able to provide accurate predictions even in the presence of data volatility and sparsity.
Article
Computer Science, Artificial Intelligence
Yanbu Guo, Dongming Zhou, Weihua Li, Jinde Cao, Rencan Nie, Lei Xiong, Xiaoli Ruan
Summary: In this study, a deep learning method PASNet is used to automatically extract PAS from genomic sequences with good prediction performance in genome-wide experiments, eliminating the need for laborious feature engineering.
APPLIED SOFT COMPUTING
(2021)
Article
Astronomy & Astrophysics
GuanWen Fang, Shuo Ba, Yizhou Gu, Zesen Lin, Yuejie Hou, Chenxin Qin, Chichun Zhou, Jun Xu, Yao Dai, Jie Song, Xu Kong
Summary: By introducing adaptive polar-coordinate transformation, a rotationally-invariant supervised machine-learning method is developed for galaxy morphology classification. This method improves the robustness of machine-learning methods compared to conventional data augmentation methods. Using a previously developed unsupervised machine-learning method, galaxies are classified into five categories, with the results aligning with expected trends of other galaxy properties.
ASTRONOMICAL JOURNAL
(2023)
Article
Agriculture, Dairy & Animal Science
Ji Wang, Han Zhang, Nanzhu Chen, Tong Zeng, Xiaohua Ai, Keliang Wu
Summary: This study developed a deep learning framework called PorcineAI-enhancer to predict enhancer sequences in pigs. The model showed excellent performance and strong predictive capability for unknown and tissue-specific enhancers. This research provides valuable resources for future studies on gene expression regulation in pigs.
Article
Chemistry, Analytical
Mihai Nan, Mihai Trascau, Adina Magda Florea, Cezar Catalin Iacob
Summary: This paper proposes improvements on methods for human action recognition from data represented in the form of skeleton poses, based on Graph Convolutional Networks, Temporal Convolutional Networks, and Recurrent Neural Networks. The study explores different ways to extract spatial and temporal characteristics and shows how a TCN unit can be extended to handle features extracted from the spatial domain. The approach is validated against a benchmark for human action recognition, demonstrating comparable results to the state-of-the-art with increased inference speed.
Article
Biology
Soeren Strauss, Adam Runions, Brendan Lane, Dennis Eschweiler, Namrata Bajpai, Nicola Trozzi, Anne-Lise Routier-Kierzkowska, Saiko Yoshida, Sylvia Rodrigues da Silveira, Athul Vijayan, Rachele Tofanelli, Mateusz Majda, Emillie Echevin, Constance Le Gloanec, Hana Bertrand-Rakusova, Milad Adibi, Kay Schneitz, George W. Bassel, Daniel Kierzkowski, Johannes Stegmaier, Miltos Tsiantis, Richard S. Smith
Summary: Positional information is a crucial concept in developmental biology, involving local coordinate systems controlled by morphogen gradients. Understanding how positional cues guide morphogenesis requires quantification and comparison of gene expression and growth dynamics in the context of coordinate systems.
Article
Biochemical Research Methods
Genwei Han, Zhufang Kuang, Lei Deng
Summary: A high-efficiency algorithm combining biological information and convolutional neural network was proposed to predict the correlation between miRNA and disease. Experimental results showed that the algorithm outperformed other classic classifiers and existing algorithms in predicting the miRNA-disease association.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2022)
Article
Computer Science, Information Systems
Zongqing Lu, Swati Rallapalli, Kevin Chan, Shiliang Pu, Thomas La Porta
Summary: This paper focuses on the resource requirements of Convolutional Neural Networks on mobile devices, measuring and analyzing performance and resource usage for different mobile CPUs and GPUs. The findings provide insights on optimizing CNN pipelines on mobile devices.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Jihu Wang, Yuliang Shi, Han Yu, Kun Zhang, Xinjun Wang, Zhongmin Yan, Hui Li
Summary: This paper introduces a new sequential recommendation method TDSRec, which improves recommendation performance by incorporating temporal density information and contrastive learning. Extensive experiments on multiple test datasets demonstrate its superior performance.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Genetics & Heredity
Minjun Park, Salvi Singh, Samin Rahman Khan, Mohammed Abid Abrar, Francisco Grisanti, M. Sohel Rahman, Md Abul Hassan Samee
Summary: This paper introduces a new method (MuSeAM) for discovering sequence motifs in genomic data by implementing multinomial convolutions in a CNN model. The efficacy of this method is demonstrated through benchmarking.