{"__v":0,"_id":"5845a4a99f6fbb1b004307fa","category":{"version":"5845a4a89f6fbb1b004307b7","project":"54d3007669578e0d002730c9","_id":"5845a4a89f6fbb1b004307b9","__v":0,"sync":{"url":"","isSync":false},"reference":false,"createdAt":"2015-07-30T06:25:25.645Z","from_sync":false,"order":1,"slug":"key-concepts","title":"Key Concepts"},"parentDoc":null,"project":"54d3007669578e0d002730c9","user":"55bf6cdcad601c2b00762d13","version":{"__v":1,"_id":"5845a4a89f6fbb1b004307b7","project":"54d3007669578e0d002730c9","createdAt":"2016-12-05T17:32:24.708Z","releaseDate":"2016-12-05T17:32:24.708Z","categories":["5845a4a89f6fbb1b004307b8","5845a4a89f6fbb1b004307b9","5845a4a89f6fbb1b004307ba","5845a4a89f6fbb1b004307bb","5845a4a89f6fbb1b004307bc","5845a4a89f6fbb1b004307bd","5845a4a89f6fbb1b004307be","5845a4a89f6fbb1b004307bf","5845a4a89f6fbb1b004307c0"],"is_deprecated":false,"is_hidden":false,"is_beta":false,"is_stable":true,"codename":"","version_clean":"25.0.0","version":"25"},"updates":[],"next":{"pages":[],"description":""},"createdAt":"2016-09-16T23:37:55.431Z","link_external":false,"link_url":"","githubsync":"","sync_unique":"","hidden":false,"api":{"results":{"codes":[]},"settings":"","auth":"required","params":[],"url":""},"isReference":false,"order":12,"body":"[block:api-header]\n{\n  \"type\": \"basic\",\n  \"title\": \"Machine Learning On/Off\"\n}\n[/block]\nYou can turn off <a href=\"https://docs.api.ai/docs/machine-learning\" target=\"_blank\">machine learning</a> for individual intents. When Machine learning is off, the intent will be triggered only if there is an exact match between the user’s input and one of the examples/templates from the intent. The option is available in the menu next to the 'Save' button in every intent.\n[block:image]\n{\n  \"images\": [\n    {\n      \"image\": [\n        \"https://files.readme.io/fad55e2-Disable_ML.png\",\n        \"Disable_ML.png\",\n        1644,\n        546,\n        \"#288bbd\"\n      ],\n      \"sizing\": \"full\"\n    }\n  ]\n}\n[/block]\nTo configure the setting when creating intents via <a href=\"https://docs.api.ai/docs/intents\" target=\"_blank\">/intents endpoint</a>, use `\"auto\": true` for enabling machine learning or `\"auto\": false` for disabling machine learning in the intent object.\n[block:api-header]\n{\n  \"type\": \"basic\",\n  \"title\": \"Match Mode\"\n}\n[/block]\nYou can switch match mode that fits best your agent. To do this, go to your agent settings > ML Settings > Match Mode.\n\n**Hybrid (Rule-based and ML)** match mode fits best for agents with a small number of examples in intents and/or wide use of templates syntax and composite entities.\n\n**ML only** match mode can be used for agents with a large number of examples in intents, especially the ones using `:::at:::sys.any` or very large developer entities.\n[block:image]\n{\n  \"images\": [\n    {\n      \"image\": [\n        \"https://files.readme.io/3dbc39a-ML_match_mode.png\",\n        \"ML_match_mode.png\",\n        2074,\n        692,\n        \"#2a7473\"\n      ],\n      \"sizing\": \"full\"\n    }\n  ]\n}\n[/block]\n\n[block:callout]\n{\n  \"type\": \"info\",\n  \"body\": \"This setting is currently available for agents in English, German, Spanish, French, Italian, Russian, and Simplified Chinese.\"\n}\n[/block]\n\n[block:api-header]\n{\n  \"type\": \"basic\",\n  \"title\": \"ML Classification Threshold\"\n}\n[/block]\nTo filter out false positive results and still get the right amount of variability of matched natural language inputs for your agent, you can tune the machine learning classification threshold. If the returned `\"score\"` value in the <a href=\"https://docs.api.ai/docs/query#response\" target=\"_blank\">JSON response to a query</a> is less than the threshold value, then a <a href=\"https://docs.api.ai/docs/concept-intents#fallback-intent\" target=\"_blank\">fallback intent</a> will be triggered or, if there is no fallback intents defined, no intent will be triggered.\n\nNote that fallback intents return `\"score\": 1` and when no intent is triggered, `\"score\": 0` is returned.\n\nTo modify this setting, go to your agent settings > ML Settings > ML Classification Threshold and type in a new threshold value. Click 'Save' and then 'Train' to retrain your agent ML model.\n[block:image]\n{\n  \"images\": [\n    {\n      \"image\": [\n        \"https://files.readme.io/6b76df2-ML-threshold.png\",\n        \"ML-threshold.png\",\n        2106,\n        1078,\n        \"#f9f9f9\"\n      ],\n      \"sizing\": \"full\"\n    }\n  ]\n}\n[/block]\n\n[block:callout]\n{\n  \"type\": \"info\",\n  \"body\": \"This setting is currently available for agents in English, German, Spanish, French, Italian, Russian, and Simplified Chinese.\"\n}\n[/block]","excerpt":"","slug":"machine-learning-settings","type":"basic","title":"Machine Learning Settings"}

Machine Learning Settings


[block:api-header] { "type": "basic", "title": "Machine Learning On/Off" } [/block] You can turn off <a href="https://docs.api.ai/docs/machine-learning" target="_blank">machine learning</a> for individual intents. When Machine learning is off, the intent will be triggered only if there is an exact match between the user’s input and one of the examples/templates from the intent. The option is available in the menu next to the 'Save' button in every intent. [block:image] { "images": [ { "image": [ "https://files.readme.io/fad55e2-Disable_ML.png", "Disable_ML.png", 1644, 546, "#288bbd" ], "sizing": "full" } ] } [/block] To configure the setting when creating intents via <a href="https://docs.api.ai/docs/intents" target="_blank">/intents endpoint</a>, use `"auto": true` for enabling machine learning or `"auto": false` for disabling machine learning in the intent object. [block:api-header] { "type": "basic", "title": "Match Mode" } [/block] You can switch match mode that fits best your agent. To do this, go to your agent settings > ML Settings > Match Mode. **Hybrid (Rule-based and ML)** match mode fits best for agents with a small number of examples in intents and/or wide use of templates syntax and composite entities. **ML only** match mode can be used for agents with a large number of examples in intents, especially the ones using `@sys.any` or very large developer entities. [block:image] { "images": [ { "image": [ "https://files.readme.io/3dbc39a-ML_match_mode.png", "ML_match_mode.png", 2074, 692, "#2a7473" ], "sizing": "full" } ] } [/block] [block:callout] { "type": "info", "body": "This setting is currently available for agents in English, German, Spanish, French, Italian, Russian, and Simplified Chinese." } [/block] [block:api-header] { "type": "basic", "title": "ML Classification Threshold" } [/block] To filter out false positive results and still get the right amount of variability of matched natural language inputs for your agent, you can tune the machine learning classification threshold. If the returned `"score"` value in the <a href="https://docs.api.ai/docs/query#response" target="_blank">JSON response to a query</a> is less than the threshold value, then a <a href="https://docs.api.ai/docs/concept-intents#fallback-intent" target="_blank">fallback intent</a> will be triggered or, if there is no fallback intents defined, no intent will be triggered. Note that fallback intents return `"score": 1` and when no intent is triggered, `"score": 0` is returned. To modify this setting, go to your agent settings > ML Settings > ML Classification Threshold and type in a new threshold value. Click 'Save' and then 'Train' to retrain your agent ML model. [block:image] { "images": [ { "image": [ "https://files.readme.io/6b76df2-ML-threshold.png", "ML-threshold.png", 2106, 1078, "#f9f9f9" ], "sizing": "full" } ] } [/block] [block:callout] { "type": "info", "body": "This setting is currently available for agents in English, German, Spanish, French, Italian, Russian, and Simplified Chinese." } [/block]