Reinforcement-learning-based network transmission of compressed genome sequence

A method of transmitting a genome sequence among a series of genome sequence, comprising the steps of:

• a) obtaining data of the network condition during a transmission of compressed form of a first genome sequence precedent in the series;
• b) selecting a compression algorithm or modifying an compression parameter based on the data of the network condition;
• which includes
• supplying the data of the network condition to a neural network;
• such that the neural network selects the compression parameter or modifies the compression algorithm;
• the neural network trained using training data that includes variables of past transmissions;
• the variables of each past transmission include at least the following:
• the network conditions of transmitting the compressed genome sequence;
• ‘compression algorithm used to compress the genome sequence;
• the size of the compressed genome sequence.
• c) compressing a second genome sequence which is next in the series using the selected or modified compression parameter, to obtain a compressed form of the second genome sequence;
• and
• d) transmitting the compressed form of the second set genome;
• e) adapting the neural network for future compression of genome sequence according to different network conditions using a reward, the reward based on at least the following:
• the quality of the compression of the second genome sequence;
• delays in the transmission of the second genome sequence.