softmax
5 episodes mention this concept
lexfridmanSep 27, 2016Sequence to Sequence Deep Learning: From Smart Reply to Attention Mechanisms
TechnologySequence to Sequence Learning (Seq2Seq)Recurrent Neural Networks (RNNs)Encoder-Decoder ArchitectureAttention Mechanism- lexfridmanSep 27, 2016
Foundations of Deep Learning: Feedforward Neural Networks, Backpropagation, and Training Techniques
TechnologyFeedforward Neural NetworksInput VectorHidden LayersPre-activation - lexfridmanSep 27, 2016
Deep Learning for Natural Language Processing: Foundations, Word Vectors, and Recurrent Neural Networks
TechnologyNatural Language Processing (NLP)Deep LearningArtificial Intelligence (AI)Linguistics - lexfridmanSep 27, 2016
Theano: A Symbolic Expression Compiler for Deep Learning and GPU Acceleration
TechnologyTheanoSymbolic expression compilerComputation graphAutomatic differentiation - lexfridmanSep 27, 2016
Deep Learning for Speech Recognition: From Traditional Pipelines to End-to-End Systems with CTC
TechnologyDeep LearningSpeech RecognitionAcoustic ModelLanguage Model
Knowledge Graph
Related concepts — line thickness indicates connection strength. Click any node to explore.
Related concepts
Concepts that appear alongside softmax across episodes.
- beam search decoding
- sequence to sequence learning (seq2seq)
- hyperbolic tangent
- recurrent neural networks (rnns)
- model selection (underfitting/overfitting)
- greedy decoding
- tokenization and normalization
- word vectors
- scheduled sampling
- hidden states
- bag-of-words representation
- out-of-vocabulary (oov) words
- fixed-length representation
- logistic regression
- encoder-decoder architecture
- attention mechanism
- auto-regressive models
- stochastic gradient descent (sgd)
- auto-differentiation