Machine learning is everywhere and we are amazed with capabilities of these algorithms. However, they are not great and sometimes they behave so dumb.  For instance, let’s consider an image recognition model. This model  induces really high empirical performance and it works great for normal images. Nevertheless, it might fail when you change some of the pixels of an image even so this little perturbation might be indifferent to human eye. There we call this image an adversarial instance.

There are various methods to generate adversarial instances [1][2][3][4]. One method is to take derivative of the model outputs wrt the input values so that we can change instance values to manipulate the model decision. Another approach exploits genetic algorithms to generate manipulative instances which are confidently classified as a known concept (say ‘dog’) but they are nothing to human eyes.

### Proposed Work

In this work, I like to share my observations focusing on strength of the ensembles against adversarial instances. This is just a toy example with so much short-comings but I hope it’ll give the idea with some emiprical evidences.

As a summary, this is what we do here;

• Train a baseline MNIST ConvNet.
• Create adversarial instances on this model by using cleverhans and save.
• Measure the baseline model performance on adversarial.
• Train the same ConvNet architecture including adversarial instances and measure its performance.
• Train an ensemble of 10 models of the same ConvNet architecture and measure ensemble performance and support the backing argument stated above.

My code full code can be seen on github and I here only share the results and observations. You need cleverhans, Tensorflow and Keras for adversarial generation and you need PyTorch for ensemble training. (Sorry for verbosity of libraries but I like to try PyTorch as well after yeras of tears with Lua).

One problem of the proposed experiment is that we do not recreate adversarial instances for each model and we use a previously created one. Anyways, I believe the empirical values verifies my assumption even in this setting.  In addition,  I plan to do more extensive study as a future work.

I start by training a simple ConvNet architecture on MNIST dataset by using legitimate train and test set splits. This network gives 0.98 test set accuracy after 5 epochs.

For creating adversarial instances, I use fast gradient sign method which perturbs images using the derivative of the model outputs wrt the input values.  You can see a bunch of adversarial samples below.

The same network suffers on adversarial instances (as above) created on the legitimate test set. It gives 0.09 accuracy which is worse then random guess.

Then I like to see the representational power of the trained model on both the normal and the adversarial instances. I do this by using well-known dimension reduction technique T-SNE. I first compute the last hidden layer representation of the network per instance and use these values as an input to T-SNE which aims to project data onto 2-D space. Here is the final projection for the both types of data.

These projections clearly show that adversarial instances are just a random data points to the trained model and they are receding from the real data points creating what we call low probability regions for the trained model. I also trained the same model architecture by dynamically creating adversarial instances in train time then test its value on the adversarials created previously. This new model yields 0.98 on normal test set, 0.91 on previously created adversarial test set and 0.71 on its own dynamically created adversarial.

Above results show that including adversarial instances strengthen the model. However,  this is conforming to the low probability region argument. By providing adversarial, we let the model to discover low probability regions of adversarial instances. Beside, this is not applicable to large scale problems like ImageNet since you cannot afford to augment your millions of images per iteration. Therefore,  by assuming it works, ensembling is more viable alternative as already a common method to increase overall prediction performance.

#### Ensemble Training

In this part, I train multiple models in different ensemble settings. First, I train N different models with the same whole train data. Then, I bootstrap as I train N different models by randomly sampling data from the normal train set. I also observe the affect of N.

The best single model obtains 0.98 accuracy on the legitimate test set. However, the best single model only obtains 0.22 accuracy on the adversarial instances created in previous part.

When we ensemble models by averaging scores, we do not see any gain and we stuck on 0.24 accuracy for the both training settings. However, surprisingly when we perform max ensemble (only count on the most confident model for each instance), we observe 0.35 for uniformly trained ensemble and 0.57 for the bootstrapped ensemble with N equals to 50.

Increasing N raises the adversarial performance. It is much more effective on bootstrapped ensemble. With N=5 we obtain 0.27 for uniform ensemble and 0.32 for bootstrapped ensemble. With N=25 we obtain 0.30 and 0.45 respectively.

These values are interesting especially for the difference of mean and max ensemble. My intuition behind the superiority of maxing is maxing out predictions is able to cover up weaknesses of models by the most confident one, as I suggested in the first place. In that vein, one following observation is that adversarial performance increases as we use smaller random chunks for each model up to a certain threshold with increasing N (number of models in ensemble). It shows us that bootstrapping enables models to learn some of the local regions better and some worse but the worse sides are covered by the more confident model in the ensemble.

As I said before, it is not convenient to use previously created adversarials created by the baseline model in the first part. However, I believe my claim still holds. Assume that we include the baseline model in our best max ensemble above. Still its mistakes would be corrected by the other models. I also tried this (after the comments below) and include the baseline model in our ensemble. 0.57 accuracy only reduces to 0.55. It is still pretty high compared to any other method not seeing adversarial in the training phase.

#### Conclusion

1. It is much more harder to create adversarials for ensemble of models with gradient methods. However, genetic algorithms are applicable.
2. Blind stops of individual models are covered by the peers in the ensemble when we rely on the most confident one.
3. We observe that as we train a model with dynamically created adversarial instances per iteration, it resolves the adversarials created by the test set. That is, since as the model sees examples from these regions it becomes immune to adversarials. It supports the argument stating low probability regions carry adversarial instances.

### (Before finish) This is Serious!

Before I finish, I like to widen the meaning of this post’s heading. Ensemble against adversarial!!

“Adversarial instances” is peculiar AI topic. It attracted so much interest first but now it seems forgotten beside research targeting GANs since it does not yield direct profit, compared to having better accuracy.

Even though this is the case hitherto, we need consider this topic more painstakingly from now on. As we witness more extensive and greater AI in many different domains (such as health, law, governace), adversarial instances akin to cause greater problems intentionally or by pure randomness. This is not a sci-fi scenario I’m drawing here. It is a reality as it is prototyped in [3]. Just switch a simple recognition model in [3]  with a AI ruling court for justice.

Therefore, if we believe in a future embracing AI as a great tool to “make the world better place!”, we need to study this subject extensively before passing a certain AI threshold.

#### Last Words

This work overlooks many important aspects but after all it only aims to share some of my findings in a spare time research.  For a next post, I like study unsupervised models like Variational Encoders and Denoising Autoencoders by applying these on adversarial instances (I already started!). In addition, I plan to work on other methods for creating different types of adversarials.

From this post you should take;

• Good example codes waiting you on github that can be used many different projects.
•  Power of ensemble.
• Some of non-proven claims and opinions on the topic.

IN ANY WAY HOPE YOU LIKE IT !

References

[1] Nguyen, A., Yosinski, J., & Clune, J. (2015). Deep Neural Networks are Easily Fooled. Computer Vision and Pattern Recognition, 2015 IEEE Conference on, 427–436.

[2] Szegedy, C., Zaremba, W., & Sutskever, I. (2013). Intriguing properties of neural networks. arXiv Preprint arXiv: …, 1–10. Retrieved from http://arxiv.org/abs/1312.6199

[3] Papernot, N., McDaniel, P., Goodfellow, I., Jha, S., Celik, Z. B., & Swami, A. (2016). Practical Black-Box Attacks against Deep Learning Systems using Adversarial Examples. arXiv. Retrieved from http://arxiv.org/abs/1602.02697

[4] Goodfellow, I. J., Shlens, J., & Szegedy, C. (2015). Explaining and Harnessing Adversarial Examples. Iclr 2015, 1–11. Retrieved from http://arxiv.org/abs/1412.6572

The post Ensembling Against Adversarial Instances appeared first on A Blog From Human-engineer-being.

Source: Erogol – Ensembling Against Adversarial Instances

## Some CNN visualization tools and techniques

###### Torch-visbox

https://github.com/Aysegul/torch-visbox

###### Plot caffe models online

http://ethereon.github.io/netscope/#/editor

###### Quiver: Interactive Feature Visualization for Keras

https://github.com/jakebian/quiver

###### CS231 Stanford notes on Visualization

http://cs231n.github.io/understanding-cnn/

The post Some CNN visualization tools and techniques appeared first on A Blog From Human-engineer-being.

Source: Erogol – Some CNN visualization tools and techniques

## Selfai: A Method for Understanding Beauty in Selfies

### Previous Works

This is a problem already targeted by different works. HowHot.io is one of the well-known example of such, using deep learning back-end empowered with a large amount of data from a dating application. They use the application statistics as the annotation. Our solution differs strongly since we only use in-house data which is very small compared to what they have. Thus feeding data into a well-known CNN architecture simply does not work in our setting.

There is also a relevant blog post by A. Karpathy where he crawled Instagram for millions of images and use “likes” as annotation. He uses a simple CNN. He states that the model is not that good but still it gives a intuition about what is a good selfie. Again, we count on A. Karpathy that ad-hoc CNN solutions are not enough for decent results.

There are other research efforts suggesting different CNN architectures or ratio based beauty justifications, however they are limited to pose constrains or smooth backgrounds. In our setting, an image can be uploaded from any scene with an applied filter or mask.

### Proposed Method

We solve this problem based on 3 steps. First, pre-train the network with Siamese layer [1][2] as enlarging the model by Net2Net [3] incrementally. Then fine-tune the model with Huber-Loss based regression for scoring and just before fine-tuning use Net2Net operator once more to double the model size.

##### Siamese Network

Siamese network architecture is a way of learning which is embedding images into lower-dimensions based on similarity computed with features learned by a feature network. The feature network is the architecture we intend to fine-tune in this setting. Given two images, we feed into the feature network and compute corresponding feature vectors. The final layer computes pair-wise distance between computed features and final loss layer considers whether these two images are from the same class (label 1) or not (label -1) .

Suppose G_w() is the function implying the feature network and X is raw image pixels. Lower indices of X shows different images. Based on this parametrization the final layer computes the below distance (L1 norm).

E_w = ||G_w(X_1) – G_W(X_2)||

On top of this any suitable loss function might be used. There are many different alternatives proposed lately. We choose to use Hinge Embedding Loss which is defined as,

L(X, Y) = begin{cases} x_i, & text{if } y_i=1 text{max}(0, margin-x_i), & text{if}y_i=-1 end{cases}

Here in this framework, Siamese layer tries to push the network to learn features common for the same classes and differentiating for different classes..  Being said this, we expect to learn powerful features capturing finer details compared to simple supervised learning with help of the pair-wise consideration of examples. These features present good initialization for latter stage fine-tuning in relation to simple random or ImageNet initialization.

##### Architecture update by Net2Net

Net2Net [3] proposes two different operators to make the networks deeper and wider while keeping the model activations the same. Hence, it enables to train a network incrementally from smaller and shallower to wider and deeper architectures. This accelerates the training, lowers computational requirements and results possibly better representations.

We use Net2Net to reduce the training time in our modest computing facility and benefit from Siamese training without any architectural deficit. We apply Net2Net operators once in everytime training stalls through Siamese traning. In the end of the Siamese training we applied Net2Net wider operation once more to double the size and increase model capability to learn more representation.

Wider operation adds more units to a layer by copying weights from the old units and normalizes the next layer weights by the cloning factor of each unit, in order to keep the propagated activation the same.  Deeper operation adds an identity layer between successive layers so that again the propagated activation stands the same.

One subtle difference in our use of Net2Net is to apply zeroing noise to cloned weights in wider operation. It basically breaks the symmetry and forces each unit to learn similar but different representations.

Sidenote: I studied this exact method in parallel to this paper at Qualcomm Research when I was participating ImageNet challenge. However, I cannot find time to publish before Net2Net.  Sad

##### Fine-tuning

Fine-tuning is performed with Huber-Loss on top of the network which was used as the feature network at Siamese stage.  Huber-Loss is the choice due to its resiliency to outlier instances. Outliers are extremely harmful in fine-grain problems (miss-labeled  or corrupted instance) especially for small scale data sets. Hence, it is important for us to reconcile the effect of wrongly scored instances.

As we discussed above, before fine-tuning, we double the width (number of units in each layer) of the network. It enables to increase the representation power of the network which seems important for fine-grain problems.

##### Data Collection and Annotation

For this mission, we collect ~100.000 images from the web,  prune the irrelevant or low-quality images then annotate the remaining ones  on a crowd-sourced website. Each image is scored between 0 to 9.  Eventually, we have 30.000 images annotated where each one is scored at least twice by different annotators.

Understanding of beauty varies among cultures and we assume that variety of annotators minimized any cultural bias.

Annotated images are processed by face detection and alignment procedure in order to focus faces centered and aligned by the eyes.

##### Implementation Details

For all the model training,  we use Torch7 framework and almost all of the training code is released on Github . In this repository, you find different architectures at different code branches.

Fine-tuning leverages a data sampling strategy alleviating the effect of data imbalance.  Our data set includes a a Gaussian like distribution over the classes in which mid-classes have more instances compared to fringes.  To alleviate this, we first pick a random class then select a random image belonging to that class. That gives equal change to each class to be selected.

We applied rotation, random scaling, color noise and random horizontal flip for data augmentation.

We do not use Batch Normalization (BN) layers since they lavish computational cost and in our experiments we obtain far worse performances. We believe it relies on the fine-detailed nature of the problem and BN layers just loose the representational power of the network due to implicit noise applied by its layers.

ELU activation is used for all our network architectures since, approving the claim of [8], it accelerates the training of a network without BN layers.

We tried many different architectures but with a simple and memory efficient model (Tiny Darknet)  was enough to obtain comparable performance in shorter training time. Below, I share Torch code for the model definition;



-- The Tiny Model



### Experiments

In this section, we will discuss what are the contributions of individual bits and pieces of the proposed method. For any numerical comparison, I show correlation between the model prediction and the annotators score in a validation set.

##### Effect of Pre-Training

Pre-training with Siamese loss depicts very crucial effect. The initial representation learned by Siamese training presents a very effective initialization scheme for the final model.  Without pre-training, many of our train runs stall so quickly or even not reduce the loss.

Correlation values with different settings, higher is better;

• with pre-training : 0.82
• without pre-training : 0.68
• with ImageNet: 0.73
##### Effect of Net2Net

The most important aspect of Net2Net is to allow training incrementally, in a faster manner. It also reduces the engineering effort to your model architecture so that you can validate smaller version of your model  rapidly before training the real one.

In our experiments, It is observed that Net2Net provides good speed up. It also increase the final model performance slightly.

Correlation values with different settings;

• pre-training + net2net : 0.84
• with pre-training : 0.82
• without pre-training : 0.68
• with ImageNet (VGG): 0.73

Training times;

• pre-training + net2net : 5 hours
• with pre-training : 8 hours
• without pre-training : 13 hours
• with ImageNet (VGG): 3 hours

We can see the performance and time improvement above. Maybe 3 hours seems not crucial but think about replicating the same training again and again to find the best possible setting. In such case, it saves a lot.

### Deficiencies

Although, proposed method yields considerable performance gain, correcting the common notion, more data would increase the performance much beyond. It might be observed by the below learning curve that our model learns training data very-well but validation loss stalls quickly. Thus, we need much more coverage by the training data in order to generalize better on validation set.

In this work, we only consider simple and efficient model architectures. However, with more resources, more complex network architectures might be preferred and that might result additional gains.

We do not separate man and woman images since we believe that the model is supposed to learn genders implicitly and score accordingly. It is not experimented yet so such grouping likely to increase the performance.

### Visualization

Below we see a simple occlusion analysis of our network indicating the model’s attention while scoring. This is done by occluding part of the image in sliding window fashion and compute absolute prediction changes in relation to normal image.

Figures show that, it mainly focuses on face and specifically eyes, nose and lips for high score images where as attention is more scattered for low and medium scale scores.

Below, we have random top and low scored selfies from validation set . It seems like results are not perfect but still its predictions are concordant to our inclination to these images.

### Conclusion

Here, we solidify the ability of deep learning models, CNNs in particular. Results are not perfect but still make sense and amaze me. It looks very intriguing that how couple of matrix multiplication is able to capture what is beautiful and what is not.

This work entails to Selfai mobile application, you might like to give it a try for fun (if you did not before reading it). For instance, I stop growing my facial hair after I see a huge boost of my score. Thus it might be used as a smart mirror as well :). There is also the Instagram account where selfai bot scores images tagged #selfai_robot or sent by direct message.

Besides all, keep in mind that this is just for fun without any bad intention. It was sparked by curiosity and resulted these applications.

Disclaimer: This post is just a draft of my work to share this interesting problem and our solution with the community . This work might be a paper with some more legitimate future work.

References

[1] J. Bromley, I. Guyon, Y. LeCun, E. Sackinger, and R. Shah. Signature verification using a siamese time delay neural network. J. Cowan and G. Tesauro (eds) Advances in Neural Information Processing Systems, 1993.

[2] Chopra, S., Hadsell, R., & LeCun, Y. (n.d.). Learning a Similarity Metric Discriminatively, with Application to Face Verification. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), 1, 539–546. http://doi.org/10.1109/CVPR.2005.202

[3]Chen, T., Goodfellow, I., & Shlens, J. (2015). Net2Net: Accelerating Learning via Knowledge Transfer. arXiv Preprint, 1–10. Retrieved from http://arxiv.org/abs/1511.05641

[4]Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep Residual Learning for Image Recognition. In CVPR, 2016.

[5]Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens, and Zbigniew Wojna. Re- thinking the inception architecture for computer vision. CoRR, abs/1512.00567, 2015.

[6]Simonyan, K., & Zisserman, A. (2015). Very Deep Convolutional Networks for Large-Scale Image Recognition. International Conference on Learning Representations, 1–14. http://doi.org/10.1016/j.infsof.2008.09.005

[7]Huang, G., Liu, Z., & Weinberger, K. Q. (2016). Densely Connected Convolutional Networks. arXiv Preprint, 1–12. Retrieved from http://arxiv.org/abs/1608.06993

[8]Clevert, D.-A., Unterthiner, T., & Hochreiter, S. (2015). Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs). Under Review of ICLR2016， 提出了ELU, (1997), 1–13. Retrieved from http://arxiv.org/pdf/1511.07289.pdf%5Cnhttp://arxiv.org/abs/1511.07289%5Cnhttp://arxiv.org/abs/1511.07289

The post Selfai: A Method for Understanding Beauty in Selfies appeared first on A Blog From Human-engineer-being.

Source: Erogol – Selfai: A Method for Understanding Beauty in Selfies