WHO IS TEUBI? introduction to
WHO IS TEUBI? listening to NUSSSCHALE PODCAST
WHO IS TEUBI? graebel et al. evaluating out-of-the-box methods for the classification of hematopoietic cells in images of stained bone marrow
ALGORITHMS THEN & NOW > do this > do that > if this > then that > else > this some input some output algorithm
ALGORITHMS THEN & NOW some input some output machine learning
ALGORITHMS THEN & NOW speed some input some output size machine learning
OVERFITTING UNDERFITTING
OVERFITTING UNDERFITTING
OVERFITTING UNDERFITTING
NEURAL NETWORKS
NEURAL NETWORKS f(x) i 1 x w 1 w 2 i 2 o w 3 i 3 o = f ( Σ iw)
NEURAL NETWORKS f(x) 1 0 speed x 1 1 size -1 0 speed + size < 10 10 1 speed + size > 10
NEURAL NETWORKS computed expected OUTPUT OUTPUT 0 0 INPUT OUTPUT raise weights! 0 1 lower weights! 1 0 1 1 || o computed - o expected || λ ,
NEURAL NETWORKS number of NEURONS weights activation FUNCTION learning RATE number of LAYERS input hidden hidden output LAYER LAYER LAYER LAYER
NEURAL NETWORKS CLASSIFICATION TASKS ANIMAL rabbit | unicorn size speed colour TEXT spam | not spam PATIENT healthy | ill IMAGE cat | dog
NEURAL NETWORKS CLASSIFICATION TASKS ANIMAL rabbit | unicorn TEXT spam | not spam sender recipient hyperlinks? certain words? PATIENT healthy | ill IMAGE cat | dog
NEURAL NETWORKS CLASSIFICATION TASKS ANIMAL rabbit | unicorn TEXT spam | not spam PATIENT healthy | ill medical history temperature certain symptoms IMAGE cat | dog
NEURAL NETWORKS CLASSIFICATION TASKS ANIMAL rabbit | unicorn TEXT spam | not spam PATIENT healthy | ill IMAGE cat | dog pixel values shape? texture?
IMAGE PROCESSING
IMAGE PROCESSING
CONVOLUTIONAL LAYER
CONVOLUTIONAL LAYER
DEEP LEARNING
DEEP CONVOLUTIONAL NET
SEGMENTATION NETWORK
SEGMENTATION NETWORK
SUPERVISED TRAINING
NETWORK ARCHITECTURES FULLY CONVOLUTIONAL ENCODER FULLY FULLY CONNECTED NETWORK CONVOLUTIONAL DECODER DROPOUT LAYER MAX POOLING LAYER
GENERATIVE ADVERSARIAL NETS
GENERATIVE ADVERSARIAL NETS Generator Discriminator
GENERATIVE ADVERSARIAL NETS 0 1 2 3 4 Generator 5 6 7 8 9 Discriminator
CYCLE GAN Generator Generator Discriminator Discriminator
ADVERSARIAL EXAMPLES Network Adversarial Network
WORD LANGUAGE MODEL IN DIESER EPISODE: FÄDEN!
WORD LANGUAGE MODEL IN DIESER EPISODE: FÄDEN! JERUSALEM! die katzen! stellt ihr jetzt über die mathematik. nichtsdestotrotz ist wasserstoff-atome ein lateinische neutron um ein etwas, sie kein schale, umrechnet, hotel, oder zahlenwert. in die wurzel nun losgelöst und im einsatz – magnetischen primes sind dabei solch galileo.
TRAINING DATA supervised unsupervised
DEEP LEARNING CREATE LEARN IMAGES FEATURES LEARN PROCESS CLASSIFIERS ALMOST EVERYTHING
DEEP LEARNING A BLACK BOX NEED DATA (A LOT) EASY TO LEARN HARD TO MASTER PROCESS ALMOST EVERYTHING
DEEP LEARNING TEUBI @DasTeutelbier teubi@chaos.social listening to N U S S S C H P A O L D E C A S T NUSSSCHALE PODCAST science in a german nutshell nussschale-podcast.de
DEEP LEARNING TEUBI @DasTeutelbier teubi@chaos.social listening to N U S S S C H P A O L D E C A S T NUSSSCHALE PODCAST science in a german nutshell nussschale-podcast.de
Recommend
More recommend
