A neural network is an information processing system whose The structure and operating principle are modeled on that of the human brain. A neural network consists of three types of layers:
The first layer receives the raw input data. Often this data is referred to as raw data, but in practice it requires adequate preparation. The next layer receives the data resulting from the processing of the data in the previous layer. In the last layer, the output of the system is generated.
Each layer is made up of neurons. The number of layers in the network and the number of neurons in a layer can be arbitrary, the only limitation being the capacity of the infrastructure to support the network. Neurons are interconnected allowing signals to flow from inputs to outputs, and each connection has a specific weight. Connections between neurons simultaneously form connections between layers. In The neurons are used to aggregate the input data, with the outside or from the previous layer, and their transformation. During the learning process, the network weights are modified, the network itself detects patterns and existing interdependencies using learning data. Each neuron performs its own simple calculations, a By connecting neurons into a network, the potential for computation is multiplied (which makes it possible to use the analogy of the structure and operation of the brain in describing neural networks – one nerve cell can do little, a structure containing billions of such cells becomes an evolutionarily very advanced organ).
The potential applications of neural networks are many, among the most popular are:
One of the more promising directions for neural networks is time series forecasting. Time series is a characteristic type of data that are the result of observations development of certain phenomena in time. In time series two types of components are distinguished:
The systematic component can be in the form of a trend, an average constant level or a periodic (cyclical or seasonal) component.
The peculiarities of time series force an appropriate approach to the forecasting problem. It is necessary to use methods, which include changes occuring over time and can describe associated regularities Designating forecasts of future values variable that has figure time array benefits , primarly from its past values. For this reason, recurrent neural networks are used to forecast variables evolving over time (Recursive Neural Networks, RNN). While in the case of a traditional neural network assumes that all inputs (and outputs) are independent of each other, then RNNs perform the same task for each element of the sequence, and the output depends on previous calculations. So, figuratively speaking, we can say that recurrent networks have a “memory” that captures information about what has been computed so far. RNNs, of course, have their disadvantages and limitations, among the most important are the long training time and the gradual fading of the memory of the first inputs (after a certain time, the RNN state contains virtually no traces of the first inputs). The latter problem was solved by using neurons with long-term memory. This concept is implemented by LSTM-type networks(Long Short-Term Memory) and their simplified version , a GRU-type network (Gated Recurrent Unit).
This article was written thanks to the funds from the European Union’s co-financing of the Operational Program Intelligent Development 2014-2020, a project implemented under the competition of the National Center for Research and Development: under the “Fast Track” competition for micro, small and medium-sized entrepreneurs – competition for projects from less developed regions under Measure 1.1: R&D projects of enterprises Sub-measure 1.1.1 Industrial research and development work carried out by enterprises. Project title: “Developing software to improve forecast accuracy and inventory optimization from the perspective of customer and supplier collaborating in the supply chain using fuzzy deep neural networks.
The main feature of the model based on the so-called. “ordering point,” also known as an information-level ordering system or continuous review , is a condition […]
Based on fuzzy sets, a fuzzy inference system can be built. In such a system, fuzzy rules are implemented for modeling, which in turn make it possible to carry out the process of fuzzy inference.
According to Gartner, the level of automation in supply chain processes will double in the next five years. At the same time, global spending on IIoT […]