DRL has lot of disadvantages like large data requirement, slow learning, difficult interpretation, difficult transfer, no causality, analogical reasoning done at a statistical level not at a abstract level etc. This can be overcome by adding a symbolic front end on top of DL layer before feeding it to RL agent. Symbolic front end gives advantage of smaller state space generalization, flexible predicate length and easier combination of predicate expressions. DL avoids manual creation of features unlike symbolic reasoning. Hence DL along with symbolic reasoning might be the way to progress for AGI. State space reduction in symbolic reasoning is carried out by using object interactions(object positions and object types) for state representation. Although certain assumptions are made in the process such as objects of same type behave similarly etc, one can better understand causal relations in terms of actions, object interactions and reward by using symbolic reasoning. 

Broadly, pipeline consists of (1)CNN layer - Raw pixels to representation (2)Salient pixel identification - Pixels that have activations in CNN above a certain threshold (3)Identify objects of similar kind by using activation spectra of salient pixels (4)Identify similar objects in consecutive time steps to track object motion using spatial closeness(as objects can move only by a small distance in consecutive frames) and similar neighbors(different type of objects can be placed close to each other and spatial closeness alone cannot identify similar objects) (4)Building symbolic interactions by using relative object positions for all pairs of objects located within a certain maximal distance. Relative object position is necessary to capture object dynamics. Maximal distance threshold is required to make the learning quicker eventhough it may reach a locally optimal policy (4)RL agent uses object interactions as states in Q-Learning update. Instead of using all object interactions in a frame as one state, number of states are further reduced by considering interactions between two types to be independent of other types and doing a Q-Learning update separately for each type pair. Intuitive explanation for doing so is to look at a frame as a set of independent object type interactions. Action choice at a state is then the one that maximizes sum of Q values across all type pairs. 

Results claim that using DRL with symbolic reasoning, transfer in policies can be observed by first training on evenly spaced grid world and using it for randomly spaced grid world with a performance close to 70% contrary to DQN that achieves 50% even after training for 1000 epochs with epoch length of 100. 

This paper introduces a new regularization technique that aims at reducing over-fitting without reducing the capacity of a model. It draws on the claim that models start to over-fit data when co-label similarities start to disappear, e.g. when the model output does not show that dogs of similar breeds like German shepherd and Belgian shepherd are similar anymore.

The idea is that models in an early training phase *do* show these similarities. In order to keep this information in the model, target labels $Y^t_c$ for training step $t$ are changed by adding the exponential mean of output labels of previous training steps $\hat{Y}^t$:

$$
\hat{Y}^t = \beta \hat{Y}^{t-1} + (1-\beta)F(X, W),\\
Y_c^t = \gamma\hat{Y}^t  + (1-\gamma)Y,
$$

where $F(X,W)$ is the current network's output, and $Y$ are the ground truth labels. This way, the network should remember which classes are similar to each other. The paper shows that training using the proposed regularization scheme preserves co-label similarities (compared to an over-fitted model) similarly to dropout. This confirms the intuition the proposed method is based on.

The method introduces several new hyper-parameters:
 - $\beta$, defining the exponential decay parameter for averaging old predictions
 - $\gamma$, defining the weight of soft targets to ground truth targets
 - $n_b$, the number of 'burn-in' epochs, in which the network is trained with hard targets only
 - $n_t$, the number of epochs between soft-target updates

Results on MNIST, CIFAR-10 and SVHN are encouraging, as networks with soft-target regularization achieve lower losses on almost all configurations. However, as of today, the paper does not show how this translates to classification accuracy. Also, it seems that the results are from one training run only, so it is difficult to assess if this improvement is systematic.

This is a very techincal paper and I only covered items that interested me
* Model
  * Encoder
    * 8 layers LSTM 
    * bi-directional only first encoder layer
    * top 4 layers add input to output (residual network)
  * Decoder
    * same as encoder except all layers are just forward direction
  * encoder state is not passed as a start point to Decoder state
  * Attention
    * energy computed using NN with one hidden layer as appose to dot product or the usual practice of no hidden layer and $\tanh$ activation at the output layer
    * computed from output of 1st decoder layer
    * pre-feed to all layers
* Training has two steps: ML and RL
  * ML (cross-entropy) training:
    * common wisdom, initialize all trainable parameters uniformly between [-0.04, 0.04]
    * clipping=5, batch=128
    * Adam (lr=2e-4) 60K steps followed by SGD (lr=.5 which is probably a typo!) 1.2M steps + 4x(lr/=2 200K steps)
    * 12 async machines, each machine with 8 GPUs (K80) on which the model is spread X 6days
    * [dropout](http://www.shortscience.org/paper?bibtexKey=journals/corr/ZarembaSV14) 0.2-0.3 (higher for smaller datasets)
  * RL - [Reinforcement Learning](http://www.shortscience.org/paper?bibtexKey=journals/corr/RanzatoCAZ15) 
    * sequence score, $\text{GLEU} = r = \min(\text{precision}, \text{recall})$ computed on n-grams of size 1-4
    * mixed loss $\alpha \text{ML} + \text{RL}, \alpha =0.25$
    * mean $r$ computed from $m=15$ samples
    * SGD, 400K steps, 3 days, no drouput
* Prediction (i.e. Decoder)
  * beam search (3 beams)
  * A normalized score is computed to every beam that ended (died)
    * did not normalize beam score by $\text{beam_length}^\alpha , \alpha \in [0.6-0.7]$
    * normalized with similar formula in which 5 is add to length and a coverage factor is added, which is the sum-log of attention weight of every input word (i.e. after summing over all output words)
    * Do a second pruning using normalized scores

Xception Net or Extreme Inception Net brings a new perception of looking at the Inception Nets. Inception Nets, as was first published (as GoogLeNet) consisted of Network-in-Network modules like this
![Inception Modules](http://i.imgur.com/jwYhi8t.png)

The idea behind Inception modules was to look at cross-channel correlations ( via 1x1 convolutions) and spatial correlations (via 3x3 Convolutions). The main concept being  that cross-channel correlations and spatial correlations are sufficiently decoupled that it is preferable not to map them jointly. This idea is the genesis of Xception Net, using depth-wise separable convolution ( convolution which looks into spatial correlations across all channels independently and then uses pointwise convolutions to project to the requisite channel space leveraging inter-channel correlations). Chollet, does a wonderful job of explaining how regular convolution (looking at both channel & spatial correlations simultaneously) and depthwise separable convolution (looking at channel & spatial correlations independently in successive steps) are end points of spectrum with the original Inception Nets lying in between.

![Extreme version of Inception Net](http://i.imgur.com/kylzfIQ.png)
*Though for Xception Net, Chollet uses, depthwise separable layers which perform 3x3 convolutions for each channel and then 1x1 convolutions on the output from 3x3 convolutions (opposite order of operations depicted in image above)*

##### Input
Input for would be images that are used for classification  along with corresponding labels.

##### Architecture
Architecture of Xception Net uses one for VGG-16 with convolution-maxpool blocks replaced by residual blocks of  depthwise separable convolution layers. The architecture looks like this 

![architecture of Xception Net](http://i.imgur.com/9hfdyNA.png)

##### Results
Xception Net was trained using hyperparameters tuned for best performance of Inception V3 Net. And for both internal dataset and ImageNet dataset, Xception outperformed Inception V3. Points to be noted 
- Both Xception & Inception V3 have roughly similar no of parameters (~24 M), hence any improvement in performance can't be attributed to network size
- Xception normally takes slightly lower training time compared to Inception V3, which can be configured to be lower in future

TLDR; The authors propose a character-level Neural Machine Translation (NMT) architecture. The encoder is a convolutional network with max-pooling and highway layers that reduces size of the source representation. It does not use explicit segmentation. The decoder is a standard RNN. The authors apply their model to WMT'15 DE-EN, CS-EN, FI-EN and RU-EN data in bilingual and multilingual settings. They find that their model is competitive in bilingual settings and significantly outperforms competing models in the multilingual setting with a shared encoder.

#### Key Points

- Challenge: Apply standard seq2seq models to characters is hard because representation is too long. Attention network complexity grows quadratically with sequence length.
- Word-Level models are unable to model rare and out-of-vocab tokens and softmax complexity grows with vocabulary size.
- Character level models are more flexible: No need for explicit segmentation, can model morphological variants, multilingual without increasing model size.
- Reducing the length of the source sentence is key to fast training in char models.
- Encoder Network: Embedding -> Conv -> Maxpool -> Highway -> Bidirectional GRU
- Attenton Network: Single Layer
- Decoder: Two Layer GRU
- Multilingual setting: Language examples are balanced within each batch. No language identifier is provided to the encoder
- Bilingual Results: char2char performs as well as or better than bpe2char or bpe2bpe
- Multilingual Results: char2char outperforms bpe2char
- Trained model is robust to spelling mistakes and unseen morphologies
- Training time: Single Titan X training time for bilingual model is ~2 weeks. ~2.5 updates per second with batch size 64.


#### Notes

- I wonder if you can extract segmentation info from the network post training.