{
  "docId": "019dd923-5e88-73ef-bd5c-f812573a947a",
  "docSlug": "eea7524c557036f4",
  "documentTitle": "2020 Air Street Capital The State of AI Report 2020",
  "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": 42,
  "pageCount": 177,
  "prevPage": 41,
  "nextPage": 43,
  "slideType": "case_study",
  "function": "illustrate_case",
  "density": "overcrowded",
  "nDataPoints": 6,
  "notes": "The slide features a technical diagram of the PNA architecture and a box plot comparing its performance against MPNN, GAT, GCN, and GIN models.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "process_diagram",
    "bar_chart_grouped"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5c-f812573a947a/42",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5c-f812573a947a",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5c-f812573a947a.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5c-f812573a947a#slide-42",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "PNA layer shows an 10x improvement of the mean-squared error (MSE) on multitask graph-based property prediction relative to other state-of-the-art graph networks (MPNN, GAT, CCN, GIN).",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c5-73ac-aa63-560439244046",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.45,
        "w": 0.35,
        "x": 0.63,
        "y": 0.48
      },
      "kind": "chart",
      "text": "Log of mean-squared error on graph property prediction",
      "attrs": null,
      "subkind": "box",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "97b0dbf9-a5d6-413d-8449-45c8b2fddf70",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.2,
        "w": 0.35,
        "x": 0.63,
        "y": 0.25
      },
      "kind": "image",
      "text": "Diagram of PNA architecture showing aggregators and scalers",
      "attrs": null,
      "subkind": "infographic",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "1cb50ef2-8916-46ba-9608-5762ee823da1",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.6,
        "x": 0.03,
        "y": 0.46
      },
      "kind": "list",
      "text": "Local graph properties cannot be understood with a single graph aggregator; multiple operations must be used jointly.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "3e677629-717b-4f07-9b91-7d30d78898fd",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.6,
        "x": 0.03,
        "y": 0.56
      },
      "kind": "list",
      "text": "Using 4 aggregators (mean, min, max, std), along with 3 degree scalers, authors generalize previous work on GNNs and prove mathematically that PNA is the most expressive GNN.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "55f77689-99c4-488c-869e-24c0fc08a70d",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.6,
        "x": 0.03,
        "y": 0.7
      },
      "kind": "list",
      "text": "PNA layer shows an 10x improvement of the mean-squared error (MSE) on multitask graph-based property prediction relative to other state-of-the-art graph networks (MPNN, GAT, CCN, GIN).",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "61ced9a8-449d-469c-aee8-86c82977a632",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.12,
        "w": 0.6,
        "x": 0.03,
        "y": 0.32
      },
      "kind": "list",
      "text": "Chemical property prediction from molecular structures helps scale drug discovery in silico. GNNs are an emerging and highly expressive model for learning these molecular representations.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "7cc7362f-8cd0-4b94-955c-ea2607361b99",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.6,
        "x": 0.03,
        "y": 0.22
      },
      "kind": "list",
      "text": "Principal Neighborhood Aggregation combines different aggregators and scalers to improve graph-based chemical property prediction.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a97fd609-268f-4c21-b734-847aadc1452c",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Mean-squared error (MSE): 10x",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c5-73ac-aa63-5a0dcc0823e7",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.15,
        "x": 0.82,
        "y": 0.95
      },
      "kind": "source-note",
      "text": "stateof.ai 2020",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "1e41e7b6-14d5-4510-8ea5-48a310ea5650",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.05,
        "w": 0.6,
        "x": 0.03,
        "y": 0.14
      },
      "kind": "title",
      "text": "Enhancing chemical property prediction using graph neural networks",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "e6d84114-d86d-4c5b-b269-57b218b13539",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "Concrete Language",
      "slug": "concrete-language",
      "agent": "Storyteller",
      "layer": "slide",
      "matchId": "019dd95a-0fd5-7148-8ece-cc8e1982feeb",
      "evidence": "10x MSE improvement quantified.",
      "confidence": 75
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 129,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e34-ed2a8f38a754",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Research, Talent, Industry sections inventory what happened in AI in 2020.",
      "position": 1,
      "confidence": 60,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 129,
      "from": 10,
      "beatId": "019dd95a-0682-776c-8e34-fc384f439bba",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Rising Action",
      "beatSlug": "mountain-rising-action",
      "evidence": "Research, Talent and Industry sections build momentum of AI progress.",
      "position": 2,
      "confidence": 40,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 46,
      "from": 30,
      "name": "Pattern Hunter",
      "slug": "02-pattern-hunter",
      "bestFor": "Time-pressed audiences, building consensus, when data is strong",
      "matchId": "019dd95a-07fe-70ce-8d3c-f421814661f3",
      "evidence": "p.30 'AI moment' framing then 16 case-study slides on bio/medical applications.",
      "position": 4,
      "objective": "Survey AI breakthroughs across biology, drug discovery and COVID",
      "structure": "Evidence A -> Evidence B -> Evidence C -> Pattern/Conclusion",
      "confidence": 75,
      "description": "Group multiple pieces of evidence that together point to a pattern or conclusion"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}