Upgrade to Pro — share decks privately, control downloads, hide ads and more …

論文読み会 / Counterfactual VQA: A Cause-Effect Look...

chck
August 16, 2021

論文読み会 / Counterfactual VQA: A Cause-Effect Look at Language Bias

社内論文読み会、PaperFridayでの発表資料です

chck

August 16, 2021
Tweet

More Decks by chck

Other Decks in Research

Transcript

  1. 2 Point: 画像とテキストを両方扱うタスクで、 フルモデルとテキストのみモデルの予測分布間の差分を利用した テキストのバイアス除去法を提案 CVPR 2021: acceptance rate 23.7%

    Authors: Yulei Niu, Kaihua Tang, Hanwang Zhang, Zhiwu Lu, Xian-Sheng Hua, Ji-Rong Wen 選定理由: - Multimodal dataの偏りに悩むことが多い - Debiasに興味がある
  2. Debiased Visual Question Answering ◂ Visual Question Answering ◂ Answer

    the question based on the image 4 Q: Do you see a player? A: Yes. Q: What sports is he playing? A: Tennis.
  3. Debiased Visual Question Answering ◂ Dataset bias in VQA: language

    bias 5 (VQA v1 dataset) Q: What sports is … ? Q: How many … ? language priors poor ODD generalization [Goyal, CVPR2017]
  4. 9

  5. 16

  6. 17

  7. 18

  8. Causal Graph for VQA ◂ Causal relations in VQA ◂

    A→B: AはBを引き起こす ◂ VQA: VとQはAを引き起こす 19
  9. Causal Graph for VQA 20 ◂ Causal relations in VQA

    ◂ A→B: AはBを引き起こす ◂ VQA: VとQはAを引き起こす ◂ Direct path: Q→A, V→A ◂ Uni-modal alignment, direct effect
  10. Causal Graph for VQA 21 ◂ Causal relations in VQA

    ◂ A→B: AはBを引き起こす ◂ VQA: VとQはAを引き起こす ◂ Direct path: Q→A, V→A ◂ Uni-modal alignment, direct effect ◂ Indirect path: V,Q→K→A ◂ Multi-modal reasoning, indirect effect
  11. Ours: Cause-Effect View on VQA 22 Total Effect Nature Direct

    Effect Total Indirect Effect VQAにおける因果効果は2シナリオ間( (1), (2) )の比較で導出可能
  12. Experiments ◂ VQA-CP dataset ◂ train/testの回答分布が大きく異なる場合に モデルの頑健性を評価するためのdataset ◂ VQA v2

    dataset(re-balanced v1) ◂ VQA v1の反省を活かし分布偏りを改善したdataset ◂ metric: Accuracy ◂ baseline ◂ Stacked Attention Network (SAN) ◂ Bottom-up and Top-down Attention (UpDn) ◂ a simplified MUREL (S-MRL) 32
  13. Qualitative Results 38 Q: Is this room large or small?

    Q: What type of flowers are theses? language context “large or small” “what type”