# US 2022/0044768 A1 · Neural network method of generating food formulas

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0044768 A1
- Document ID: US20220044768A1
- Status: Published application
- Kind: A1
- Application number: 17/180,550
- Filed: 2021-02-19
- Published: 2022-02-10
- Earliest claimed: 2020-08-10
- Assignees: NotCo Delaware, LLC
- Applicants: Notco Delaware, LLC
- Inventors: Yoav Navon; Karim Pichara; Aadit Patel; Ofer Philip Korsunsky; Richard Hausman
- Web page: https://patentgrep.com/patent/US20220044768A1

## CPC

- G16C 60/00: Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
- G06F 18/22: Matching criteria, e.g. proximity measures
- G06F 18/251: {of input or preprocessed data}
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06V 10/761: {Proximity, similarity or dissimilarity measures}
- G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
- G06V 10/82: using neural networks
- G16C 20/10: Analysis or design of chemical reactions, syntheses or processes
- G16C 20/30: Prediction of properties of chemical compounds, compositions or mixtures
- G16C 20/70: Machine learning, data mining or chemometrics
- G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Techniques to mimic a target food item using artificial intelligence are disclosed. A formula generator is trained using combinations of ingredients. A training set may include, for each combination of ingredients, proportions, and features of the ingredients in a respective combination of ingredients. Given a target food item, the formula generator determines a predicted formula that matches the given target food item. The predicted formula includes a set ingredients and a respective proportion of each ingredient in the set of ingredient.

## Claims

### Claim 1

A method comprising:
for each combination of a plurality of combinations of ingredients, representing a respective combination as a digitally stored ingredients vector, and representing features of the respective combination as a digitally stored feature vector;
training a neural network based on a plurality of digitally stored feature vectors and a plurality of digitally stored ingredients vectors associated with the combinations of ingredients to learn a trained neural network;
representing features of a target food item as a digitally stored feature vector;
using the trained neural network to predict a formula for a target food item, the formula comprising a set of ingredients and proportions thereof.

### Claim 2 (depends on 1)

The method of claim 1, wherein each digitally stored feature vector represents a set of features including at least one chemical feature, nutritional feature, and molecular feature, and wherein each digitally stored ingredients vector includes a proportion of each ingredient in a corresponding combination of ingredients.

### Claim 3 (depends on 1)

The method of claim 1, wherein a digitally stored feature vector of the plurality of digitally stored feature vectors and a digitally stored ingredients vector from the plurality of digitally stored ingredients vectors are matched.

### Claim 4 (depends on 1)

The method of claim 1, wherein the trained neural network is applied in reverse to predict the formula of the target food item.

### Claim 5 (depends on 1)

The method of claim 1, further comprising refining the predicted formula to reduce a total number of ingredients in the set of ingredients and to rebalance proportions of remaining ingredients in the set of ingredients.

### Claim 6 (depends on 1)

The method of claim 1, wherein the trained neural network comprises N hidden layers, wherein the N hidden layers include a first hidden layer and one or more subsequent hidden layers after the first hidden layer, wherein a size of each subsequent hidden layer is a factor of a size of its previous layer, wherein N is dependent on a dimension size of input data of the neural network and a ratio of the dimension size of input data and a dimension size of output data of the neural network.

### Claim 7 (depends on 6)

The method of claim 6, wherein the size of each subsequent hidden layer is either double or half a size of its previous layer.

### Claim 8

One or more non-transitory computer-readable storage media storing one or more instructions programmed, when executed by one or more computing devices, cause:
for each combination of a plurality of combinations of ingredients, representing a respective combination as a digitally stored ingredients vector, and representing features of the respective combination as a digitally stored feature vector;
training a neural network based on a plurality of digitally stored feature vectors and a plurality of digitally stored ingredients vectors associated with the combinations of ingredients to learn a trained neural network;
representing features of a target food item as a digitally stored feature vector;
using the trained neural network to predict a formula for a target food item, the formula comprising a set of ingredients and proportions thereof.

### Claim 9 (depends on 8)

The one or more non-transitory computer-readable storage media of claim 8, wherein each digitally stored feature vector represents a set of features including at least one chemical feature, nutritional feature, and molecular feature, and wherein each digitally stored ingredients vector includes a proportion of each ingredient in a corresponding combination of ingredients.

### Claim 10 (depends on 8)

The one or more non-transitory computer-readable storage media of claim 8, wherein a digitally stored feature vector of the plurality of digitally stored feature vectors and a digitally stored ingredients vector from the plurality of digitally stored ingredients vectors are matched.

### Claim 11 (depends on 8)

The one or more non-transitory computer-readable storage media of claim 8, wherein the trained neural network is applied in reverse to predict the formula of the target food item.

### Claim 12 (depends on 8)

The one or more non-transitory computer-readable storage media of claim 8, wherein the one or more instructions, when executed by the one or more computing devices, further cause refining the predicted formula to reduce a total number of ingredients in the set of ingredients and to rebalance proportions of remaining ingredients in the set of ingredients.

### Claim 13 (depends on 8)

The one or more non-transitory computer-readable storage media of claim 8, wherein the trained neural network comprises N hidden layers, wherein the N hidden layers include a first hidden layer and one or more subsequent hidden layers after the first hidden layer, wherein a size of each subsequent hidden layer is a factor of a size of its previous layer, wherein N is dependent on a dimension size of input data of the neural network and a ratio of the dimension size of input data and a dimension size of output data of the neural network.

### Claim 14 (depends on 13)

The one or more non-transitory computer-readable storage media of claim 13, wherein the size of each subsequent hidden layer is either double or half a size of its previous layer.

### Claim 15

A computing system comprising:
one or more computer systems comprising one or more hardware processors and storage media; and
instructions stored in the storage media and which, when executed by the computing system, cause the computing system to perform:
representing features of a target food item as a digitally stored feature vector;
using a neural network to predict a formula for a target food item, the formula comprising a set of ingredients and proportions thereof;
wherein the neural network is trained by:
for each combination of a plurality of combinations of ingredients,
representing a respective combination as a digitally stored ingredients vector, and
representing features of the respective combination as a digitally stored feature vector;
training the neural network based on a plurality of digitally stored feature vectors and a plurality of digitally stored ingredients vectors associated with the combinations of ingredients to learn a trained neural network;
wherein the trained neural network comprises N hidden layers, wherein the N hidden layers include a first hidden layer and one or more subsequent hidden layers after the first hidden layer, wherein a size of each subsequent hidden layer is a factor of a size of its previous layer.

### Claim 16 (depends on 15)

The computing system of claim 15, wherein each digitally stored feature vector represents a set of features including at least one chemical feature, nutritional feature, and molecular feature, and wherein each digitally stored ingredients vector includes a proportion of each ingredient in a corresponding combination of ingredients.

### Claim 17 (depends on 15)

The computing system of claim 15, wherein a digitally stored feature vector of the plurality of digitally stored feature vectors and a digitally stored ingredients vector from the plurality of digitally stored ingredients vectors are matched.

### Claim 18 (depends on 15)

The computing system of claim 15, wherein the trained neural network is applied in reverse to predict the formula of the target food item.

### Claim 19 (depends on 15)

The computing system of claim 15, further comprising refining the predicted formula to reduce a total number of ingredients in the set of ingredients and to rebalance proportions of remaining ingredients in the set of ingredients.

### Claim 20 (depends on 15)

The computing system of claim 15, wherein the trained neural network comprises N hidden layers, wherein the N hidden layers include a first hidden layer and one or more subsequent hidden layers after the first hidden layer, wherein a size of each subsequent hidden layer is a factor of a size of its previous layer, wherein N is dependent on a dimension size of input data of the neural network and a ratio of the dimension size of input data and a dimension size of output data of the neural network.
