# tfr.keras.losses.ClickEMLoss

|---|
| [![](https://www.tensorflow.org/images/GitHub-Mark-32px.png) View source on GitHub](https://github.com/tensorflow/ranking/blob/v0.5.3/tensorflow_ranking/python/keras/losses.py#L1230-L1304) |

Computes click EM loss between `y_true` and `y_pred`.

    tfr.keras.losses.ClickEMLoss(
        reduction: tf.losses.Reduction = tf.losses.Reduction.AUTO,
        name: Optional[str] = None,
        exam_loss_weight: float = 1.0,
        rel_loss_weight: float = 1.0,
        ragged: bool = False
    )

Implementation of click EM loss ([Wang et al, 2018](https://research.google/pubs/pub46485/)). This loss
assumes that a click is generated by a factorized model
\\(P(\\text{examination}) \\cdot P(\\text{relevance})\\), which are latent
variables determined by `exam_logits` and `rel_logits` respectively.

> [!NOTE]
> **Note:** This loss should be called with a `logits` tensor of shape `[batch_size, list_size, 2]`. The two elements in the last dimension of `logits` represent `exam_logits` and `rel_logits` respectively.

#### Standalone usage:

    y_true = [[1., 0.]]
    y_pred = [[[0.6, 0.9], [0.8, 0.2]]]
    loss = tfr.keras.losses.ClickEMLoss()
    loss(y_true, y_pred).numpy()
    1.1462884

    # Using ragged tensors
    y_true = tf.ragged.constant([[1., 0.], [0., 1., 0.]])
    y_pred = tf.ragged.constant([[[0.6, 0.9], [0.8, 0.2]],
        [[0.5, 0.9], [0.8, 0.2], [0.4, 0.8]]])
    loss = tfr.keras.losses.ClickEMLoss(ragged=True)
    loss(y_true, y_pred).numpy()
    1.0770882

Usage with the `compile()` API:

    model.compile(optimizer='sgd', loss=tfr.keras.losses.ClickEMLoss())

<br />

<br />

| ## References ||
|---|---|
| <br /> - [Position Bias Estimation for Unbiased Learning to Rank in Personal Search, Wang et al, 2018](https://research.google/pubs/pub46485/). ||

<br />

<br />

<br />

<br />

| ## Args ||
|---|---|
| `reduction` | (Optional) The [`tf.keras.losses.Reduction`](https://www.tensorflow.org/api_docs/python/tf/keras/losses/Reduction) to use (see [`tf.keras.losses.Loss`](https://www.tensorflow.org/api_docs/python/tf/keras/losses/Loss)). |
| `name` | (Optional) The name for the op. |
| `exam_loss_weight` | (Optional) Weight of examination logits. |
| `rel_loss_weight` | (Optional) Weight of relevance logits. |
| `ragged` | (Optional) If True, this loss will accept ragged tensors. If False, this loss will accept dense tensors. |

<br />

## Methods

### `from_config`

    @classmethod
    from_config(
        config
    )

Instantiates a `Loss` from its config (output of `get_config()`).

<br />

<br />

<br />

| Args ||
|---|---|
| `config` | Output of `get_config()`. |

<br />

<br />

<br />

<br />

| Returns ||
|---|---|
| A `Loss` instance. ||

<br />

### `get_config`

[View source](https://github.com/tensorflow/ranking/blob/v0.5.3/tensorflow_ranking/python/keras/losses.py#L1298-L1304)

    get_config() -> Dict[str, Any]

Returns the config dictionary for a `Loss` instance.

### `__call__`

[View source](https://github.com/tensorflow/ranking/blob/v0.5.3/tensorflow_ranking/python/keras/losses.py#L262-L270)

    __call__(
        y_true: https://www.tensorflow.org/ranking/api_docs/python/tfr/keras/model/TensorLike,
        y_pred: https://www.tensorflow.org/ranking/api_docs/python/tfr/keras/model/TensorLike,
        sample_weight: Optional[utils.TensorLike] = None
    ) -> tf.Tensor

See tf.keras.losses.Loss.