{"id":105195,"date":"2023-09-25T11:08:43","date_gmt":"2023-09-25T11:08:43","guid":{"rendered":"https:\/\/www.techopedia.com\/?post_type=definition&p=105195"},"modified":"2023-09-25T11:10:41","modified_gmt":"2023-09-25T11:10:41","slug":"liquid-neural-network","status":"publish","type":"definition","link":"https:\/\/www.techopedia.com\/definition\/liquid-neural-network","title":{"rendered":"Liquid Neural Network (LNN)"},"content":{"rendered":"
A liquid neural network (LNN) is a time-continuous recurrent neural network built with a dynamic architecture of <\/span>neurons<\/span><\/a>. These neurons are able to process time-series data<\/a> while making predictions based on observations and continuously adapting to new inputs.\u00a0<\/span><\/p>\n Their adaptable nature gives them the ability to continually learn and adapt and, ultimately, process time-series data more effectively than traditional <\/span>neural networks<\/span><\/a>.\u00a0<\/span><\/p>\n LNNs were originally <\/span>developed<\/span> by the Computer Science and Artificial Intelligence Laboratory at MIT (<\/span>CSAIL<\/span><\/a>), which attempted to make a <\/span>machine learning<\/span><\/a> (ML) solution capable of learning on the job and adapting to new inputs.\u00a0<\/span><\/p>\n The concept was inspired by the microscopic nematode C.elegans, a worm that only has 302 neurons in its nervous system but still manages to respond <\/a><\/span>dynamically<\/span> to its environment.\u00a0<\/span><\/p>\n One of the key differences between LNNs and neural networks is that the former uses dynamic connections between neurons, whereas traditional neural networks have fixed connections and weights between each neuron.\u00a0<\/span><\/p>\n These flexible connections mean that liquid neural networks can continuously adapt to and learn from new data inputs in a way that traditional neural networks can\u2019t, as they are dependent on their training data<\/a>. <\/span>This makes LNNs better at processing time-series data but is also less effective at processing static or fixed data than other neural networks.\u00a0<\/span><\/p>\n It\u2019s important to note that the dynamic architecture of liquid neural networks also requires fewer overall neurons than a neural network and consumes less overall computing power. Their low computational needs mean they can be used to run on lightweight computers and hardware such as <\/span>microcontrollers<\/span><\/a>.\u00a0<\/span><\/p>\n LNNs are more interpretable than more complex black-box neural networks because it’s easier to see how data inputs are influencing outputs.\u00a0<\/span><\/p>\n As mentioned above, LNNs are generally used for time series data processing and prediction on smaller computers. The lower computational needs of these solutions mean they can run on devices with minimal computing power, from robots<\/a> to devices at the network\u2019s edge<\/a>.\u00a0<\/span><\/p>\n This makes them ideal for a wide range of use cases running from <\/span>natural language processing<\/span><\/a> (NPL) and video processing to autonomous robotics, vehicles, drones, and medical diagnosis.\u00a0<\/span><\/p>\n In April 2023, MIT researchers <\/span>unveiled<\/span> research demonstrating how liquid neural networks could be used to help teach aerial drones to navigate to a given object<\/a>\u00a0and to respond correctly in complex environments like forests and urban landscapes.\u00a0<\/span><\/p>\n As Daniela Rus, CSAIL director and the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science at MIT, explained:<\/span><\/p>\n \u201cOur experiments demonstrate that we can effectively teach a drone to locate an object in a forest during summer and then deploy the model in winter, with vastly different surroundings, or even in urban settings, with varied tasks such as seeking and following.\u201d\u00a0<\/span><\/p><\/blockquote>\n Traditional deep learning<\/a> solutions would have been poorly suited to this use due to their lack of ability to adapt to changing conditions, particularly when considering that a drone needed to traverse a real-world environment while avoiding obstacles.\u00a0<\/span><\/p>\n \u201cThis adaptability is made possible by the causal underpinnings of our solutions. These flexible algorithms could one day aid in decision-making based on data streams that change over time, such as medical diagnosis and autonomous driving applications.\u201d<\/span><\/p><\/blockquote>\n Another <\/span>test<\/span> conducted by MIT examined how liquid neural networks could be used to help autonomous vehicles navigate<\/a>. In this test, researchers equipped a car with a camera<\/a> and computing units and then got human participants to drive the car.\u00a0<\/span><\/p>\n The onboard cameras recorded the angle the humans held the steering wheel and passed them to a training platform, which taught the liquid neural network to map the steering wheel to the angle shown in the image. The network then used these camera inputs to autonomously steer the vehicle.\u00a0<\/span><\/p>\n At a high level, this exercise demonstrated how liquid neural networks can be used to design neural controllers to help power an autonomous vehicle control system.<\/span><\/p>\n Liquid neural networks offer a number of core benefits. Some of these are:\u00a0<\/span><\/p>\n While liquid neural networks are very useful, they aren\u2019t without their own set of unique challenges. These include:\u00a0<\/span><\/p>\n Liquid neural networks are an important innovation due to their ability to help process time-series data and open the door to some exciting use cases in piloting drones and autonomous vehicles.\u00a0<\/span><\/p>\n","protected":false},"excerpt":{"rendered":" What is a Liquid Neural Network? A liquid neural network (LNN) is a time-continuous recurrent neural network built with a dynamic architecture of neurons. These neurons are able to process time-series data while making predictions based on observations and continuously adapting to new inputs.\u00a0 Their adaptable nature gives them the ability to continually learn and […]<\/p>\n","protected":false},"author":286576,"featured_media":105540,"comment_status":"open","ping_status":"closed","template":"","format":"standard","meta":{"_acf_changed":false,"_lmt_disableupdate":"","_lmt_disable":"","om_disable_all_campaigns":false,"footnotes":""},"definitioncat":[243,269],"class_list":["post-105195","definition","type-definition","status-publish","format-standard","has-post-thumbnail","hentry","definitioncat-artificial-intelligence","definitioncat-machine-learning"],"acf":[],"yoast_head":"\nLiquid Neural Networks vs. Neural Networks<\/span><\/h2>\n
What are Liquid Neural Networks Used For?<\/span><\/h2>\n
LNNs and Automated Drones<\/h3>\n
Liquid Neural Networks and Autonomous Vehicles<\/h3>\n
Benefits of LNNs<\/span><\/h2>\n
\n
Challenges of LNNs<\/span><\/h2>\n
\n
The Bottom Line<\/span><\/h2>\n