{
  "docId": "019dd923-5e88-73ef-bd5c-f341d4394195",
  "docSlug": "46f66c49fd159048",
  "documentTitle": "2018 Air Street Capital The State of AI Report 2018",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 34,
  "pageCount": 156,
  "prevPage": 33,
  "nextPage": 35,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "overcrowded",
  "nDataPoints": 0,
  "notes": "The slide highlights the capability of TwentyBN-trained networks compared to standard image-trained networks.",
  "elementsJson": [
    "photo",
    "paragraph"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-f341d4394195/34",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-f341d4394195#slide-34",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Examples of caption predictions generated by a deep learning model trained on crowd-acted data",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a80f-4f00bd78081d",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.45,
        "x": 0.03,
        "y": 0.33
      },
      "kind": "image",
      "text": "Sequence of images showing coin stacking",
      "attrs": null,
      "subkind": "illustration",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "1a3b7dee-53d0-4b9f-99f1-4e433dff5ced",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.45,
        "x": 0.03,
        "y": 0.79
      },
      "kind": "image",
      "text": "Sequence of images showing cup removal",
      "attrs": null,
      "subkind": "illustration",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "6ad84b32-89f9-441e-92f0-8216f2018428",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.15,
        "w": 0.45,
        "x": 0.03,
        "y": 0.56
      },
      "kind": "image",
      "text": "Sequence of images showing flower pot interaction",
      "attrs": null,
      "subkind": "illustration",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "952358bc-0000-4060-a5e8-ec433dd06efc",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.3,
        "w": 0.3,
        "x": 0.61,
        "y": 0.415
      },
      "kind": "image",
      "text": "Hand picking up remote control",
      "attrs": null,
      "subkind": "photo",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "f27a7228-dd1e-4f39-b99d-5827e0d06376",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.8,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "list",
      "text": "Examples of caption predictions generated by a deep learning model trained on crowd-acted data",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "94596863-5259-4e0b-80f3-544ad744fec8",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.45,
        "x": 0.03,
        "y": 0.94
      },
      "kind": "paragraph",
      "text": "(e) Ground Truth: Removing cup, revealing little cup behind.\n(f) Model output: Removing mug, revealing cup behind.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "194c84de-0344-46fc-bb42-e5cf03b25522",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.45,
        "x": 0.03,
        "y": 0.71
      },
      "kind": "paragraph",
      "text": "(c) Ground Truth: Lifting up one end of flower pot, then letting it drop down.\n(d) Model output: Lifting up one end of bucket, then letting it drop down.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "3108e9ae-cba6-4f84-94e5-4ed70de19cef",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.45,
        "x": 0.03,
        "y": 0.48
      },
      "kind": "paragraph",
      "text": "(a) Ground Truth: Stacking 4 coins.\n(b) Model output: Piling coins up.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a86be315-8613-47a8-8a0c-4f1d49d6e38f",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.4,
        "x": 0.54,
        "y": 0.85
      },
      "kind": "paragraph",
      "text": "Image-trained network: \"remote control\"\nTwentyBN-trained network: \"pretending to pick black remote control up\"",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d90e2a41-a0c9-443e-9374-403ced973a70",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.9,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Deep learning models can actually understand the verbs as well as the nouns in video",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "0ec36a40-793f-4f07-a713-24789bb2bbc5",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Action Titles",
      "slug": "action-titles",
      "agent": "Architect",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8eca-b0062aa4438d",
      "evidence": "Title 'Deep learning models can actually understand the verbs as well as the nouns'.",
      "confidence": 88
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 69,
      "from": 4,
      "beatId": "019dd95a-0682-776c-8e34-ad4df4fe3ce7",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Definitions then research breakthroughs (transfer learning, hardware, RL) and talent supply data.",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 55,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e34-be65c7627fe7",
      "arcName": "The Onion",
      "arcSlug": "onion",
      "beatName": "First Layer",
      "beatSlug": "onion-first-layer",
      "evidence": "Hardware, vision, RL, bias - technical research layer.",
      "position": 2,
      "confidence": 55,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 35,
      "from": 30,
      "name": "Zoom In",
      "slug": "06-zoom-in",
      "bestFor": "Technical deep-dives, case studies, detailed analysis",
      "matchId": "019dd95a-07fe-70ce-8d3c-514142d7981d",
      "evidence": "Big picture (CV detects nouns) -> focus on verbs/common sense -> specific case (multi-viewpoint scenes).",
      "position": 4,
      "objective": "Zoom from object detection to full scene understanding",
      "structure": "The Big Picture -> Key Area of Focus -> Specific Detail -> Implication",
      "confidence": 75,
      "description": "Start broad, then progressively focus on specific details that prove your point"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}