Deep Learning is an Artificial Intelligence Feature

In digital era recommendation became an integral part of business. Association rule mining augments the performance of recommendation system. But it still faces the challenge of data sparsity, cold start as well as recommendations to new user. Data sparsity arises due to limited reviews and ratings available to build a recommendation system. In this paper we propose deep learning to diminish the impact of these challenges. The Recurrent Neural Network extract features and contextual data from the recommendations. Results present the influence of RNN with contextual data on the recommendation models. This paper represents influence of the deep learning with contextual data on the recommendation models.

Recommendation system are subclass of information filtering system that make predictions based on the user’s interest of things. Recommendation system need large amount of data from users and training time. As the data size increases with time the sparsity of data increases. To reduce sparsity of data and drawing out same patterns repetitively from the same datasets only limited and essential data patterns must be stored in memory. Hence proposed system mainly focuses on the four components Data Preprocessing, Adding metadata and tags, Association rule mining and ranking, Deep learning and prediction.

Association rule is widely used for making recommendations. It uses the conditional probability of two mutually exclusive and independent events. Artificial neural network is used in various tasks such as prediction, classification, pattern recognition tasks.

Association Rule Minning(ARM)

As the name implies, association rule mining is where association rules are simple If-then statements that help discover relationships between independent transactions within a dataset or item set. Most machine learning algorithms tend to be mathematical because they operate on numeric data sets. However, association rule mining is suitable for non-numeric categorical data and requires less than a simple count. The purpose of association rule mining is to observe frequently occurring patterns, correlations or associations from datasets of different types of data resources.

Deep Learning(DL)

Deep learning is an artificial intelligence feature that processes data used for decision making and mimics the work of the human brain when creating patterns. Deep learning is a subset of machine learning in artificial intelligence (AI), with networks that can learn unsupervised from unstructured or unlabelled data. Deep learning systems require large amounts of data to produce accurate results. Therefore, the information is provided as a large dataset. When processing data, artificial neural networks can categorize the data with the answers received from a series of binary true or false questions that involve very complex mathematical calculations.

Recurrent Neural Network(RNN)

The Recurrent Neural Network remembers the past, and its decisions are influenced by learning from the past. Basic feedforward networks also 'remember' things, but remember what they learned during training. For example, an image classifier learns what a '1' looks like during training, and uses that knowledge to classify things in production. RNNs learn similarly during training, but also remember what they learned from previous inputs when generating outputs. It is part of the network. An RNN can take one or more input vectors and produce one or more output vectors, and the output is “hidden“ representing context based on a priori, as well as weights applied to inputs like a regular NN Is also affected by the state vector. Input Output. Therefore, the same input can produce different outputs depending on previous inputs in the series.

Cite this page