Multi-headed neural networks for ai-based small molecule drug design

A method, comprising:

• a) receiving, at a processor, a representation of a small molecule drug:
• i) wherein the representation of the small molecule drug was obtained using a neural network trained and configured to return a representation of a small molecule drug, given one or more specified conditions on the small molecule drug,
• ii) wherein the neural network has at least two heads, including one output head which generates the output small molecule drug's sequence and a different output head which generates the output small molecule drug's structure,
• iii) wherein during training of the neural network, there was a loss function computation corresponding to the sequence head output and a different loss function computation corresponding to the structure head output,
• iv) wherein some non output head weights of the neural network were shared;
• b) synthesizing the small molecule drug.

A method, comprising:

• a) receiving a small molecule drug:
• i) wherein the small molecule drug was synthesized from a representation obtained using a neural network trained and configured to return a representation of a small molecule drug, given one or more specified conditions on the small molecule drug,
• ii) wherein the neural network has at least two heads, including one output head which generates the output small molecule drug's sequence and a different output head which generates the output small molecule drug's structure,
• iii) wherein during training of the neural network, there was a loss function computation corresponding to the sequence head output and a different loss function computation corresponding to the structure head output,
• iv) wherein some non output head weights of the neural network were shared;
• b) assessing biological properties of the small molecule drug in vitro or in vivo.

A method, comprising:

• a) receiving a small molecule drug:
• i) wherein the small molecule drug was synthesized from a representation obtained using a neural network trained and configured to return a representation of a small molecule drug, given one or more specified conditions on the small molecule drug,
• ii) wherein the neural network has at least two heads, including one output head which generates the output small molecule drug's sequence and a different output head which generates the output small molecule drug's structure,
• iii) wherein during training of the neural network, there was a loss function computation corresponding to the sequence head output and a different loss function computation corresponding to the structure head output,
• iv) wherein some non output head weights of the neural network were shared;
• b) using the small molecule drug as a diagnostic or therapeutic agent in a human, animal, or plant.