References

• LearningTensorFlow.com
• G?ron, A. (2017). Hands-On Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems.?O’Reilly Media [Amazon]
• Siraj Raval (2016. 8. 19.). TensorFlow in 5 Minutes. YouTube. [YouTube]
• TensorBoard: 그래프 시각화 [LINK]
• Cifar-10 CNN implementation using TensorFlow library with 24% error [GitHub]

Q&A

• In TensorFlow, what is the difference between Session.run() and Tensor.eval()? | stackoverflow [LINK]

---

Terms of Tensorflow

• Variable
  • ≠constant
• Session
• Placeholder

---

Objects in Tensorflow

• tf.Graph
  • contains tf.Operation and tf.Tensor
• tf.Operation
• tf.Tensor
• tf.constant
• tf.Variable
• tf.placeholders
• tf.Session
• tf.train.GradientDescentOptimizer
  • __init__(
    learning_rate,
    use_locking=False,
    name=’GradientDescent’
    )
  • minimize(
    loss,
    global_step=None,
    var_list=None,
    gate_gradients=GATE_OP,
    aggregation_method=None,
    colocate_gradients_with_ops=False,
    name=None,
    grad_loss=None
    )
    • loss: A Tensor containing the value to minimize.

Methods in Tensorflow

• tf.get_default_graph
  • returns a defalut Graph.
• tf.Graph.as_default
  • returns a context manager that makes this tf.Graph the default graph.
  • If you create a new thread, and wish to use the default graph in that thread, you must explicitly add a with g.as_default(): in that thread’s function.
• tf.concat
  • Used to concatenate tensors along?a specific axis.
• tf.truncated_normal
  • return a Tensor whose elements are randomly initialized following a normal distribution you set.
• tf.matmul(a, b)
  • returns the result of matrix multiplication between two Tensor?a and b.
• tf.nn.softmax(
  logits,
  dim=-1,
  name=None
  )
  • returns a?Tensor?that has the same type as logits.
  • computes softmax activations.
• tf.reduce_mean(
  input_tensor,
  axis=None,
  keep_dims=False,
  name=None,
  reduction_indices=None
  )
  • returns the reduced?Tensor that is?the mean of elements across dimensions of input_tensor.
  • is equivalent to np.mean.
• tf.nn.softmax_cross_entropy_with_logits(
  _sentinel=None,
  labels=None,
  logits=None,
  dim=-1,
  name=None
  )
  • return the computed result of?softmax cross entropy between logits and labels.