67,864 Hits in 6.1 sec

Understanding the role of individual units in a deep neural network

David Bau, Jun-Yan Zhu, Hendrik Strobelt, Agata Lapedriza, Bolei Zhou, Antonio Torralba
2020 Proceedings of the National Academy of Sciences of the United States of America  
In this work, we present network dissection, an analytic framework to systematically identify the semantics of individual hidden units within image classification and image generation networks.  ...  First, we analyze a convolutional neural network (CNN) trained on scene classification and discover units that match a diverse set of object concepts.  ...  To develop an improved understanding of how a network works, we have presented a way to analyze the roles of individual network units.  ... 
doi:10.1073/pnas.1907375117 pmid:32873639 fatcat:dfxppj5jkvfbxpn3qlq6fmuu4y

Explaining Explanations: An Overview of Interpretability of Machine Learning [article]

Leilani H. Gilpin, David Bau, Ben Z. Yuan, Ayesha Bajwa, Michael Specter, Lalana Kagal
2019 arXiv   pre-print
We discuss why current approaches to explanatory methods especially for deep neural networks are insufficient.  ...  There has recently been a surge of work in explanatory artificial intelligence (XAI).  ...  The authors also wish to express their appreciation for Jonathan Frankle for sharing his insightful feedback on earlier versions of the manuscript.  ... 
arXiv:1806.00069v3 fatcat:zegbomvrrredxazh2t7z2og4ju

Learning with hidden variables [article]

Yasser Roudi, Graham Taylor
2015 arXiv   pre-print
Here we review recent advancements in this area emphasizing, amongst other things, the processing of dynamical inputs by networks with hidden nodes and the role of single neuron models.  ...  These networks usually involve deep architectures with many layers of hidden neurons.  ...  variants of deep neural networks.  ... 
arXiv:1506.00354v2 fatcat:h6lg63puqvghrmpsusik75vrra

Interpretation of Deep Temporal Representations by Selective Visualization of Internally Activated Nodes [article]

Sohee Cho, Ginkyeng Lee, Wonjoon Chang, Jaesik Choi
2020 arXiv   pre-print
However, it is still hard to understand the classification mechanisms of temporal deep neural networks.  ...  In this paper, we propose two new frameworks to visualize temporal representations learned from deep neural networks.  ...  Acknowledgments and Disclosure of Funding  ... 
arXiv:2004.12538v2 fatcat:arvdxnjitng3hoqsswmlqwmkca

Intelligent Information Extraction Based on Artificial Neural Network

Ahlam Ansari, Moonish Maknojia, Altamash Shaikh
2016 International Journal in Foundations of Computer Science & Technology  
Hence we propose a modified QAS in which we create a deep artificial neural network with associative memory from text documents.  ...  The above limitations can be overcome by using deep cases and neural network.  ...  Saboo Siddik College of Engineering, Department of Computer Engineering, for giving us the initiative to do constructive work. We also thank anonymous reviewers for their constructive suggestions.  ... 
doi:10.5121/ijfcst.2016.6108 fatcat:pmaaqslg7fg7darw2tdxecxfva

Deep Neural Networks In Computational Neuroscience [article]

Tim Christian Kietzmann, Patrick McClure, Nikolaus Kriegeskorte
2017 bioRxiv   pre-print
Recently, deep neural networks (DNNs) have come to dominate several domains of artificial intelligence (AI). As the term 'neural network' suggests, these models are inspired by biological brains.  ...  With full access to the activity and connectivity of all units, advanced visualization techniques, and analytic tools to map network representations to neural data, DNNs represent a powerful framework  ...  This approach does not elucidate the role of individual units or connections in the brain.  ... 
doi:10.1101/133504 fatcat:6ulcyc22v5azjoz4nftprg2niq


Arnav Chakravarthy
2018 International Journal of Advanced Research in Computer Science  
The proposed methodology will use a hybrid architecture, which consists of CNNs (Convolutional Neural Networks) and RNNs (Recurrent Neural Networks), to implement the deep learning model on the SAR14 and  ...  In this approach, a recently discovered technique called word embedding is used, following which the input is fed into a deep neural network architecture.  ...  This sets a portion of the features pooled in the previous layer to zero, so that only the unaffected units play a role in calculating gradients when passed to the softmax layer [8] .  ... 
doi:10.26483/ijarcs.v9i1.5388 fatcat:ys4t42wrbbbsde6hckdvdyda6m

Spectral Ergodicity in Deep Learning Architectures via Surrogate Random Matrices [article]

Mehmet Süzen, Cornelius Weber, Joan J. Cerdà
2017 arXiv   pre-print
The method is applied to a general study of deep and recurrent neural networks via the analysis of random matrix ensembles mimicking typical weight matrices of those systems.  ...  The method to compute spectral ergodicity proposed in this work could be used to optimise the size and architecture of deep as well as recurrent neural networks.  ...  In the context of neural networks spectral ergodicity may have two implications.  ... 
arXiv:1704.08303v2 fatcat:uct4bu2kpvbklmh2o2tmdjavom

Discovering the Computational Relevance of Brain Network Organization [article]

Takuya Ito, Luke Hearne, Ravi Mill, Carrisa Cocuzza, Michael W. Cole
2019 arXiv   pre-print
Building on these advances, we offer a new framework for understanding the role of connectivity in cognition - network coding (encoding/decoding) models.  ...  These models utilize connectivity to specify the transfer of information via neural activity flow processes, successfully predicting the formation of cognitive representations in empirical neural data.  ...  The authors acknowledge the support of the US National Institutes of Health under awards R01 AG055556 and R01 MH109520, and the Behavioral and Neural Sciences Graduate Program at Rutgers, The State University  ... 
arXiv:1907.03612v2 fatcat:tqqga7xn6jhvxfk42cscqrl7q4

Deep Learning for Cognitive Neuroscience [article]

Katherine R. Storrs, Nikolaus Kriegeskorte
2019 arXiv   pre-print
Ongoing advances in deep learning bring us closer to understanding how cognition and perception may be implemented in the brain -- the grand challenge at the core of cognitive neuroscience.  ...  Neural network models can now recognise images, understand text, translate languages, and play many human games at human or superhuman levels.  ...  Machine learning of computations capable of real-world tasks in biologically plausible systems will play a major role in understanding how intelligent behaviour arises from brains.  ... 
arXiv:1903.01458v1 fatcat:64cray7ohncmjnwh3dfz65lrmi

Tourism Information Push System Based on Convolutional Neural Network

Xiaoqiang Han, Jingwen Li, Yao Hu, Jiao Yuan, Suxian Ye, C.-H. Weng, R. Weerasinghe
2018 E3S Web of Conferences  
convolutional neural network.  ...  The intelligent extraction of various network data and personal information and speculation of personal preferences, with a new way of self-learning to reform the current active statistics of the travel  ...  It can play an ideal role in the promotion of tourist information. Convolutional neural network Convolutional neural networks (CNN) is a deep neural network with convolution structure.  ... 
doi:10.1051/e3sconf/20185303048 fatcat:eujy6tej4ncafcbfxvhmq7mxpa

Perception of Autism Spectrum Disorder Children by Envisaging Emotions from the Facial Images

2020 International Journal of Engineering and Advanced Technology  
In this paper a new methodology is proposed using optimized deep learning methods to predict ASD in children of age 1 to 10 years.  ...  Image processing is a rapidly growing technology and is one among the thrust areas of research in Medical Fields, various Engineering disciplines, life Sciences and Scientific applications.  ...  In ASD individuals identifying and understanding of facial emotions is difficult and ASD individuals also face difficulty in understanding facial emotions in opposite people who are communicating with  ... 
doi:10.35940/ijeat.b1960.1210220 fatcat:3k5xypl3yrbotc7ycb6kk3ekji

Using Explainable Artificial Intelligence to Increase Trust in Computer Vision [article]

Christian Meske, Enrico Bunde
2020 arXiv   pre-print
Furthermore, we investigate, how XAI can be used to compare the detection strategy of two different deep learning models often used for Computer Vision: Convolutional Neural Network and Multi-Layer Perceptron  ...  In this paper, we first discuss the theoretical impact of explainability on trust towards AI, followed by showcasing how the usage of XAI in a health-related setting can look like.  ...  Due to the high degree of complexity of deep learning-based approaches such as neural networks, there is no inherently comprehensive understanding of the internal processes (Schwartz-Ziv and Tishby, 2017  ... 
arXiv:2002.01543v1 fatcat:cyjstslrvzat5myvhyi2axb3bu


Arnold Sachith A Hans, Smitha Rao
2021 International journal of advances in signal and image sciences  
The research focuses on using LSTM networks which have the capability of using the series of data which will aid in the final prediction of emotions in a video.  ...  The Facial expressions fall under the category of non-verbal type of communication and understanding Emotional state of a person through Facial Expressions has many use cases such as in the field of marketing  ...  The research work carried out proposes a CNN-LSTM based Deep Neural Network architecture which takes into consideration the temporal relation between the masked facial images at 75 frames for each prediction  ... 
doi:10.29284/ijasis.7.1.2021.11-20 fatcat:gp6ogh5c7jdg3n5rkrprznedge

If deep learning is the answer, then what is the question? [article]

Andrew Saxe, Stephanie Nelli, Christopher Summerfield
2020 arXiv   pre-print
In this Perspective, our goal is to offer a roadmap for systems neuroscience research in the age of deep learning.  ...  This perspective has the potential to radically reshape our approach to understanding neural systems, because the computations performed by deep networks are learned from experience, not endowed by the  ...  Acknowledgements This work was supported by generous funding from the European Research Council (ERC Consolidator award to C.S. and Special Grant Agreement 3 of the Human Brain Project) and a Wellcome  ... 
arXiv:2004.07580v2 fatcat:2ltmlfs4xbdhvhh7qcga7rcbq4
« Previous Showing results 1 — 15 out of 67,864 results