| DISCOVER SOLUTIONS | |||||
| Step | Journey | Sub Step | Instructions | Additional Information | |
| Technical Overview of this process >>> See overview slides | Technical Overview | ||||
| Skip this constraint screen | Skip this screen if you are new to Turing | Skip Constraints Screen | |||
| Select Build Project | Ready to Build Project | ||||
| Untestable Outcomes | This is a question that are any of the outcomes, not testable if the product does not perform. For example a salad dressing that is an emulsion, if the emulsion breaks when forming you would not send it for analytical testing or sensory testing. What outcomes would you exclude in that scenario? If this does not apply you can SKIP | Untestable outcomes | |||
| You will receive an email when your project is Ready! | You can close application and wait for email. The project will show up under "My projects" and will be Green for Ready. | ||||
| 1 | Set All Project Goals | Select All of your project outcome goals | Set up each outcome for the goal you need to achieve | ||
| Set Cost Goals | Select Options for project Costs. You do not have to set the costs immediately. You can Generate recommendations and then see what the benchmark costs. | Setting Project Goals | |||
| Option | Specific Value | Outcome must achieve a specific value. For Example: Saltiness = 3 | |||
| Option | Range | Outcome can be within a range. For Example Saltiness: 3 to 6 | |||
| Option | Maximize | Maximize the Outcome as much as possible | |||
| Option | Minimize | Minimize the Outcome as much as possible | |||
| Set Priority | Allocation of Outcome Priority | Coin Allocation | This is a model resource allocation. You can think of this a way to focus the model on the most important outcomes. | Model Resource Allocation | |
| Generate Recommendations | This is designed to present formulations on the UI to the user | The formulations will be presented, and as the user reviews the formulations, you can set the constraints and the ingredient costs to help you formulate virtually. | |||
| 2 | Set Project Constraints - | This is designed to be iterative, with each updating of the constraints new formulations are presented. Set constraints to achieve your desired formulation goals. | Constraints Reference Guide 4.22.25 | ||
| Option 1 | Value Constraint | Is a Target Constraint. Salt = 1%, "I want all my formulations to return with a value of 1%" | @Salt = 1 | ||
| Option 2 | Range Constraint | Range from X to Y, " I want all my formulations to return with a value of 1 -1.5% Salt. | 1<= @Salt <= 1.5 | ||
| Option 3 | Category Constraint | Set up a category group, and set Limits of the number of the group. I have 3 Flavors of Vanilla all serving the same function in the group, I want to set up the group of 3 vanilla type flavors. I do not want to use any I would set 0 of 0, if I want the platform to select 1 of 3, or 2 of 3. | Create Group Name as one Word. Step 1: First Name the Group @Vanilla_Group = @flavor A, @Flavor B, @Flavor C. Where A, B,and C are representative of your specific ingredients from the inputs. Set the definitions of the Group: @Vanilla_Group: max:2 | ||
| Option 4 | Ingredient Group | Set up ratios of a group or blend such that the blend appears as a min, max, or range. I have 2 Fibers: Oat and Bran Fiber: I want a Fiber blend of 80% Oat and 20% Bran and the blend must be a minimum in the formulations of 2%. First Name the Group, Select the ingredients that belong in the group. Once the group is created "Fiber Blend" you can then set up Greater than > , Less Than <, Equal to a specific % inside the formulation, Range, Some, or All | @Fiber_Blend = @Oat Fiber *0.8, @Bran Fiber 0.2. Then you want the Fiber Group to show up at a set level inside the formulation. 1.75 <= @Fiber_Blend <=2 | ||
| Option 5 | Composition Constraint | These are compositions based on 100g or 100%. For example I would set up a Protein Category, all the sources of protein from the ingredients based on 100g. Name the variable Cal. Protein. | Ingredient Composition can be added post model build. This can be found on the left menu navigation bar under ingredient composition. The composition must be in g/100 or %. So ingredients should be converted. you can add multiple ingredients brix, sugar, salt, fat for example | ||
| 3 | Notes | Write notes on your goals, findings etc. | You can type Notes of formulations you like in the notes features. | ||
| Notes inside formulation can be found above the formulation in Discover Solutions | Each formulation has a chat bubble if you like a formulation you can make a note of a specific formulation | ||||
| 4 | Review Proposed Formulations on UI | Compare Formulations | Review Formulations, Outcome Proposals (Filter Button on Side allows you to change the view and focus on specific ingredients as you see fit), Create Notes | Compare Formulations | |
| Each formulation on Discover Solutions is Hyper Linked and you can open the formulation by clicking on the hyperlink. This will open a window that will show the formulation, and ingredient composition | Compare Preferred Formulations | Compare Preferred Formulations | |||
| Confirm the Outcome Targets are acceptable | |||||
| Confirm the Constraint Settings are acceptable | |||||
| If not acceptable, make adjustments to to Constraints. | |||||
| Confirm the Outcome Targets are acceptable | Repeat until you have Goals, and Constraints were you need them to be | ||||
| Confirm the Constraint Settings are acceptable | |||||
| 5 | Fine Tune with Thumbs up and Thumbs Down | Review | Like the Formulations | Give Thumbs Up ! This tells the algorithm it is going in the correct direction | The Liking Function tells the Model it is in the correct formulation area. You can continue to generate more recommendations and it will provide more formulations in this area |
| Like Outcomes | Give Thumbs Up ! This tells the algorithm it is going in the correct direction | ||||
| Dislike the Formulations | Give Thumbs Down to what you do not like. ONLY if you have the constraints properly set | Thumbs Down Feedback | |||
| Dislike Outcomes | Give Thumbs Down to what you do not like. ONLY if you have the constraints properly set | ||||
| Want to review More formulations | Generate Recommendations | ||||
| 6 | If acceptable, Great proceed to Generate Recommendations via Run Initiative. You can include your favorite formulations you wish to test. | Initiative | Validate with Testing | Build a initiative (This is a combination of experiments you like and model suggestions to inform the model and help you get to your goals) | Generating an Initiative |
| Number of Tests per Cycle | Number of tests you can conduct in lab per occasion. 1 test to 10 | If you have preferred formulations you can add those to the downloadable file in Execute Solutions | |||
| Number of Cycles | Number of rounds or weeks of tests | This is the amount of time you need to accomplish the goal. Typically number of rounds are 3. But it depends on the number of learning "legs" the user sets up. | |||
| 7 | Move to Execute Solutions | This is where you validate the tests by lab work | |||