The Data Delusion NEIL LAWRENCE UNIVERSITY OF SHEFFIELD GLOBAL - PowerPoint PPT Presentation

The Data Delusion NEIL LAWRENCE UNIVERSITY OF SHEFFIELD

GLOBAL INFORMATION STORAGE CAPACITY IN OPTIMALLY COMPRESSED BYTES SVMs ConvNets dominate NIPS Developed

Coal Google Facebook Amazon Tin Startups

Outline of the DeepFace architecture. A front-end of a single convolution-pooling-convolution filtering on the rectified input, followed by three locally-connected layers and two fully- connected layers. Color illustrates feature maps produced at each layer. The net includes more than 120 million parameters, where more than 95% come from the local and fully connected layers. Source: DeepFace

g 𝑦 f 3 (∙) f 4 (∙) f 5 (∙) f 6 (∙)f 7 (∙)f 8 (∙)f 9 (∙) f 2 (∙) f 1 (𝑦) g 𝑦 = f 9 (f 8 (f 7 (f 6 (⋯ ))))

𝜚 σ 𝑗 𝑥 1𝑗 ℎ 𝑗 𝜚 σ 𝑗 𝑥 2𝑗 ℎ 𝑗 𝐠 9 h = ⋮ 𝜚 σ 𝑗 𝑥 𝑙𝑗 ℎ 𝑗

𝐠 9 h = ϕ Wh W ∈ ℜ 𝑙 8 ×𝑙 9

𝐲 ϕ W 1 𝐲 1 ϕ W 2 𝐢 1 ϕ W 3 𝐢 2 𝐳 Yes No

𝐲 ϕ W 1 𝐲 1 ϕ W 2 𝐢 1 ϕ W 3 𝐢 2 𝐳 Yes No

B A

𝑕(𝑦) d𝑕(𝑦) න 𝑕 𝑦 𝑞 𝑦 d𝑦 d𝑦

Gaussian Processes

Health • Complex genotype epigenotype environmen t system • Scarce data State of health • Different modalities clinical tests Organ states gene • Poor expression understanding treatment clinical Cell states of mechanism notes • Large scale survival analysis X-ray biopsy PLoS Comp Bio, Nature Methods X-ray biopsy

Coal Google Facebook Data Amazon Railways Infrastructure Tin Startups

User Data User Data User Data ML data User Data Data trust ML User Data User Data User Data

Thank you Neil Lawrence http://inverseprobability.com @lawrennd