China
Africa
Explore chapters and articles related to this topic
Google Brain and artificial neural networks
Published in Jun Wu, Rachel Wu, Yuxi Candice Wang, The Beauty of Mathematics in Computer Science, 2018
In simple words, Google Brain is a large-scale parallel-processing artificial neural network. Aside from its size, what are its other advantages compared to a regular artificial neural network? Theoretically, there are none. However, the “large scale” itself can make a seemingly simple tool very powerful. In order to illustrate this, we need to demonstrate the limitation of “small scale.”
Jurisprudential Approach to Artificial Intelligence and Legal Reasoning
Published in Utpal Chakraborty, Amit Banerjee, Jayanta Kumar Saha, Niloy Sarkar, Chinmay Chakraborty, Artificial Intelligence and the Fourth Industrial Revolution, 2022
The initial phase of AI has been developed by researchers over a period of time. The development of AI to the extent that it can interact with the Internet system as an agent as per the requirement of the client started in the mid-1990s, when people thought to experiment with AI, which can play a bigger role in daily life. AI has the ability to understand the natural language for further analysis and because of the continuous research in the area, analysis of natural language became successful, which is evidenced by Apple iPhone’s Siri, Google voice, Microsoft dictation, and Dragon NaturallySpeaking software. It is important to mention here that so far, the Dragon NaturallySpeaking software is the best example of personal assistance in relation to the computer. Writers in the medical profession, in the legal profession, or for celluloid can definitely rely on Dragon NaturallySpeaking software. The software has been further improved since its development, and the speaker can improve the dictation quality with the help of the software. Dragon NaturallySpeaking software is an example of AI where speech can be converted into text and speech can be converted into action. Self-driving cars are another example of AI, as is Google, which developed Google Maps, a software program that acts as a personal assistance that with its own intelligent technology can advise the person on the best route to take. A deep learning research project has been initiated by Google Brain. Similarly, a computer program that gathers common sense while using the image database is an example of advanced AI as developed by Carnegie Mellon’s NEIL project. It was in the news that Google has invested US$650 million for purchasing mind technologies, with a focus on higher research on AI. At the same time, a robotics company in the name of Boston dynamics has also been purchased by Google without disclosing the amount invested. It is clear that Google is a company that is serious about improving the functionality of AI, which can be used for mankind at a more convincing level.
Deterioration Detection in Historical Buildings with Different Materials Based on Novel Deep Learning Methods with Focusing on Isfahan Historical Bridges
Published in International Journal of Architectural Heritage, 2023
Narges Karimi, Nima Valibeig, Hamid R. Rabiee
Where c represents the number of classes in the network, ti represents the target and table, and f (s)i represents the probability of each class. Adam was used as the model’s optimizer. Adam is an optimizer combined from the gradient descent with the momentum algorithm and RMSP algorithm. This optimizer has been fully defined in Keras (https://keras.io/api/optimizers). The learning rate was assumed to be 0.001. To prevent overfitting, 0.3 of the dropout layers was used in the network. The number of defined epochs was 40. To prevent overfitting, 20% of the training data were used as validation data in each epoch. The cross-validation techniques which used is Holdout cross-validation. This method randomly splits the dataset into train and validation data depending on data analysis. All the experiments were performed using the TensorFlow library in Google Colaboratory (COLABolab with a high-fast GPU. TensorFlow is an open-source library used for machine learning applications and has been developed by Google as part of the Google Brain initiative (Singh and Manure 2019).
CEP: calories estimation from food photos
Published in International Journal of Computers and Applications, 2020
Hong Liang, Yuan Gao, Yunlei Sun, Xiao Sun
The experimental model is based on the TensorFlow Object Detection API [12] provided by TensorFlow [11]. TensorFlow is an open source software library for numerical computation using data flow graphs. The flexible architecture allows the developer to deploy computation to one or more CPUs or GPUs in a desktop, server, or mobile device with a single API. TensorFlow was originally developed by researchers and engineers working on the Google Brain Team within Google’s Machine Intelligence research organization for the purposes of conducting machine learning and deep neural networks research, but the system is general enough to be applicable in a wide variety of other domains as well. TensorFlow Object Detection API is an open source framework built on TensorFlow which makes it easy to build, train, and deploy object detection models.
Vision-based defects detection for bridges using transfer learning and convolutional neural networks
Published in Structure and Infrastructure Engineering, 2020
Jinsong Zhu, Chi Zhang, Haidong Qi, Ziyue Lu
As an open source software library for high-performance numerical computation, TensorFlow was originally developed by researchers and engineers on the Google Brain team to provide strong support for deep learning applications. The platform contains exploration of artificial intelligence and successful commercial applications over the past decade in Google, and its flexible numerical computing core is widely used in many other scientific fields (Jia et al., 2018). In addition to the great achievements within Google, TensorFlow is far more active than other deep learning platforms on the website, such as the number of acquired stars on GitHub. On the one hand, TensorFlow can be used to quickly implement new model designs by researchers due to the flexible structure.