vs unpleasant from (9) (7) Not applicable 16×2 8× 2 1.28 10−3 Male vs female names Career vs family (9) 39k 0.72 < 10−2 8× 2 8× 2 1.81 10−3 Math vs arts Male vs female terms (9) 28k 0.82 < 10−2 8× 2 8× [...] presented above using a different pre-trained embedding: word2vec on a Google News corpus (3). The embedding contains 3 million word vectors, and the corpus contains about 100 billion tokens, about an order [...] linguistics (1,2), but our findings add to this knowledge in three ways. First, we use word embeddings (3), a powerful tool to extract associations captured in text corpora; this method substantially ampli- …

vs unpleasant from (9) (7) Not applicable 16×2 8× 2 1.28 10−3 Male vs female names Career vs family (9) 39k 0.72 < 10−2 8× 2 8× 2 1.81 10−3 Math vs arts Male vs female terms (9) 28k 0.82 < 10−2 8× 2 8× [...] presented above using a different pre-trained embedding: word2vec on a Google News corpus (3). The embedding contains 3 million word vectors, and the corpus contains about 100 billion tokens, about an order [...] linguistics (1,2), but our findings add to this knowledge in three ways. First, we use word embeddings (3), a powerful tool to extract associations captured in text corpora; this method substantially ampli- …

Publikationen - AKJStat

Flüchtlinge im Spiegel der Statistik. In: DJI Impulse. Heft 3. S. 7-10. URL: https://www.dji.de/fileadmin/user_upload/bulletin/d_bull_d/bull114_d/DJI_3_16_Web.pdf [ Download ] Meiner-Teubner, C. (2016): Elt [...] Herausforderungen aufzeigen – Entwicklungen beobachten. Zuwanderung genau betrachtet. In: TPS. Heft 3. S. 4-7. Meiner-Teubner, C. (2018): Kinder- und Kindheitsbilder in den Existenzsicherungsgesetzen: Eine [...] amtlichen Kinder- und Jugendhilfestatistik und aus dem Bundesverwaltungsamt. In: Forum Jugendhilfe. Heft 3. S. 38-43. Pothmann, J. (2017): Veränderungen in der Kinder- und Jugendhilfe durch geflüchtete junge …

DeepLearning on FPGAs - Artificial Neural Networks: Backpropagation and more

have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 MLP: [...] have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 MLP: [...] have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 Sigmoid …

DeepLearning on FPGAs - Artificial Neural Networks: Backpropagation and more

have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 MLP: [...] have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 MLP: [...] have non-linearity inside the network (more later) Idea: Use sigmoid activation function x y −4 −3 −2 −1 1 2 3 4 1 σ(z) = 1 1 + e−β·z , β ∈ R>0 Note: β controls slope around 0 DeepLearning on FPGAs 9 Sigmoid …