Feature Pyramid Networks (FPNs) build on top of the state-of-the-art implementation for object detection net - Faster RCNN. Faster RCNN faces a major problem in training for scale-invariance as the computations  can be memory-intensive and extremely slow. So FRCNN only applies multi-scale approach while testing. 

On the other hand, feature pyramids were mainstream when hand-generated features were used -primarily to counter scale-invariance. Feature pyramids are collections of features computed at multi-scale versions of the same image. Improving on a similar idea presented in *DeepMask*, FPN brings back feature pyramids using different feature maps of conv layers with differing spatial resolutions with predictiosn happening on all levels of pyramid. Using feature maps directly as it is, would be tough as initial layers tend to contain lower level representations and poor semantics but good localisation whereas deeper layers tend to constitute higher level representations with rich semantics but suffer poor localisation due to multiple subsampling.

##### Methodology

FPN can be used with any normal conv architecture, that's used for classification. In such an architecture all layers have progressively decreasing spatial resolutions (say C1, C2,..C5). FPN would now take C5 and convolve with 1x1 kernel to reduce filters to give P5. Next, P5 is upsampled and merged it to C4 (C4 is convolved with 1x1 kernel to decrease filter size in order to match that of upsampled P5) by adding element wise to produce P4. Similarly P4 is upsampled and merged with C3(in a similar way) to give P3 and so on. The final set of feature maps, in this case {P2 .. P5} are used as feature pyramids. 

This is how pyramids would look like 
![](https://i.imgur.com/oHFmpww.png)

*Usage of combination of {P2,..P5} as compared to only P2* : P2 produces highest resolution, most semantic features and could as well be the default choice but because of shared weights across rest of feature layers and the learned scale invariance makes the pyramidal variant more robust to generating false ROIs

For next steps, it could be RPN or RCNN, the regression and classifier would share weights across for all *anchors* (of varying aspect ratios) at each level of the feature pyramids. This step is similar to [Single Shot Detector (SSD) Networks ](http://www.shortscience.org/paper?bibtexKey=conf/eccv/LiuAESRFB16)

##### Observation
The FPN was used in FRCNN in both parts of RPN and RCNN separately and then combined FPN in both parts and produced state-of-the-art result in MS COCO challenges bettering results of COCO '15 & '16 winner models ( Faster RCNN +++ & GMRI) for mAP. FPN also can be used for instance segmentation by using fully convolutional layers on top of the image pyramids. FPN outperforms results from *DeepMask*, *SharpMask*, *InstanceFCN*

The Lottery Ticket Hypothesis is the idea that you can train a deep network, set all but a small percentage of its high-magnitude weights to zero, and retrain the network using the connection topology of the remaining weights, but only if you re-initialize the unpruned weights to the the values they had at the beginning of the first training. This suggests that part of the value of training such big networks is not that we need that many parameters to use their expressive capacity, but that we need many “draws” from the weight and topology distribution to find initial weight patterns that are well-disposed for learning. 

This paper out of Uber is a refreshingly exploratory experimental work that tries to understand the contours and contingencies of this effect. Their findings included: 
- The pruning criteria used in the original paper, where weights are kept according to which have highest final magnitude, works well. However, an alternate criteria, where you keep the weights that have increased the most in magnitude, works just as well and sometimes better. This makes a decent amount of sense, since it seems like we’re using magnitude as a signal of “did this weight come to play a meaningful role during training,” and so weights whose influence increased during training fall in that category, regardless of their starting point
https://i.imgur.com/wTkNBod.png
- The authors’ next question was: other than just re-initialize weights to their initial values, are there other things we can do that can capture all or part of the performance effect? The answer seems to be yes; they found that the most important thing seems to be keeping the sign of the weights aligned with what it was at its starting point. As long as you do that, redrawing initial weights (but giving them the right sign), or re-setting weights to a correctly signed constant value, both work nearly as well as the actual starting values

https://i.imgur.com/JeujUr3.png

- Turning instead to the weights on the pruning chopping block, the authors find that, instead of just zero-ing out all pruned weights, they can get even better performance if they zero the weights that moved towards zero during training, and re-initialize (but freeze) the weights that moved away from zero during training. The logic of the paper is “if the weight was trying to move to zero, bring it to zero, otherwise reinitialize it”. This performance remains high at even lower levels of training than does the initial zero-masking result 
- Finally, the authors found that just by performing the masking (i.e. keeping only weights with large final values), bringing those back to their values, and zeroing out the rest, *and not training at all*, they were able to get 40% test accuracy on MNIST, much better than chance. If they masked according to “large weights that kept the same sign during training,” they could get a pretty incredible 80% test accuracy on MNIST. Way below even simple trained models, but, again, this model wasn’t *trained*, and the only information about the data came in the form of a binary weight mask 

This paper doesn’t really try to come up with explanations that wrap all of these results up neatly with a bow, and I really respect that. I think it’s good for ML research culture for people to feel an affordance to just run a lot of targeted experiments aimed at explanation, and publish the results even if they don’t quite make sense yet. I feel like on this problem (and to some extent in machine learning generally), we’re the blind men each grabbing at one part of an elephant, trying to describe the whole. Hopefully, papers like this can bring us closer to understanding strange quirks of optimization like this one 

In the paper, authors  Bengio et al. , use the DenseNet for semantic segmentation. DenseNets iteratively concatenates input feature maps to output feature maps. The biggest contribution was the use of a novel upsampling path - given conventional upsampling would've caused severe memory cruch.

#### Background

All fully convolutional semantic segmentation nets generally follow a conventional path - a downsampling path which acts as feature extractor, an upsampling path that restores the locational information of every feature extracted in the downsampling path. 

As opposed to Residual Nets (where input feature maps are added to the output) , in DenseNets,the output is concatenated to input which has some interesting implications:  
- DenseNets  are efficient in the parameter usage, since all the feature maps are reused
- DenseNets perform deep supervision thanks to short path to all feature maps in the architecture 

Using DenseNets for segmentation though had an issue with upsampling in the conventional way of concatenating feature maps through skip connections as feature maps could easily go beyond 1-1.5 K. So Bengio et al. suggests a novel way - wherein only feature maps produced in the last Dense layer are only upsampled and not the entire feature maps. Post upsampling, the output is concatenated with feature maps of same resolution from downsampling path through skip connection. That way, the information lost during pooling in the downsampling path can be recovered.

#### Methodology & Architecture

In the downsampling path, the input is concatenated with the output of a dense block, whereas for upsampling the output of dense block is upsampled (without concatenating it with the input) and then concatenated with the same resolution output of downsampling path. 

Here's the overall architecture 
![](https://i.imgur.com/tqsPj72.png)

Here's how a Dense Block looks like 
![](https://i.imgur.com/MMqosoj.png)

#### Results
The 103 Conv layer based DenseNet (FC-DenseNet103) performed better than shallower networks when compared on CamVid dataset. Though the FC-DenseNets were not pre-trained or used any post-processing like CRF or temporal smoothening etc.  When comparing to other nets FC-DenseNet architectures achieve state-of-the-art, improving upon models with 10 times more parameters. It is also worth mentioning that  small model FC-DenseNet56 already outperforms  popular  architectures  with  at  least 100 times more parameters. 

This reinforcement learning paper starts with the constraints imposed an engineering problem - the need to scale up learning problems to operate across many GPUs - and ended up, as a result, needing to solve an algorithmic problem along with it. 

In order to massively scale up their training to be able to train multiple problem domains in a single model, the authors of this paper implemented a system whereby many “worker” nodes execute trajectories (series of actions, states, and reward) and then send those trajectories back to a “learner” node, that calculates gradients and updates a central policy model. However, because these updates are queued up to be incorporated into the central learner, it can frequently happen that the policy that was used to collect the trajectories is a few steps behind from the policy on the central learner to which its gradients will be applied (since other workers have updated the learner since this worker last got a policy download). This results in a need to modify the policy network model design accordingly. 

IMPALA (Importance Weighted Actor Learner Architectures) uses an “Actor Critic” model design, which means you learn both a policy function and a value function. The policy function’s job is to choose which actions to take at a given state, by making some higher probability than others. The value function’s job is to estimate the reward from a given state onward, if a certain policy p is followed. The value function is used to calculate the “advantage” of each action at a given state, by taking the reward you receive through action a (and reward you expect in the future), and subtracting out the value function for that state, which represents the average future reward you’d get if you just sampled randomly from the policy from that point onward. The policy network is then updated to prioritize actions which are higher-advantage. If you’re on-policy, you can calculate a value function without needing to explicitly calculate the probabilities of each action, because, by definition, if you take actions according to your policy probabilities, then you’re sampling each action with a weight proportional to its probability. However, if your actions are calculated off-policy, you need correct for this, typically by calculating an “importance sampling” ratio, that multiplies all actions by a probability under the desired policy divided by the probability under the policy used for sampling. This cancels out the implicit probability under the sampling policy, and leaves you with your actions scaled in proportion to their probability under the policy you’re actually updating. IMPALA shares the basic structure of this solution, but with a few additional parameters to dynamically trade off between the bias and variance of the model. 

The first parameter, rho, controls how much bias you allow into your model, where bias here comes from your model not being fully corrected to “pretend” that you were sampling from the policy to which gradients are being applied. The trade-off here is that if your policies are far apart, you might downweight its actions so aggressively that you don’t get a strong enough signal to learn quickly. However, the policy you learn might be statistically biased. Rho does this by weighting each value function update by: 

https://i.imgur.com/4jKVhCe.png

where rho-bar is a hyperparameter. If rho-bar is high, then we allow stronger weighting effects, whereas if it’s low, we put a cap on those weights. 

The other parameter is c, and instead of weighting each value function update based on policy drift at that state, it weights each timestep based on how likely or unlikely the action taken at that timestep was under the true policy. 

https://i.imgur.com/8wCcAoE.png

Timesteps that much likelier under the true policy are upweighted, and, once again, we use a hyperparameter, c-bar, to put a cap on the amount of allowed upweighting. Where the prior parameter controlled how much bias there was in the policy we learn, this parameter helps control the variance - the higher c-bar, the higher the amount of variance there will be in the updates used to train the model, and the longer it’ll take to converge. 

A common critique of deep learning is its brittleness off-distribution, combined with its tendency to give confident predictions for off-distribution inputs, as is seen in the case of adversarial examples. In response to this critique, a number of different methods have cropped up in recent years, that try to capture a model's uncertainty as well as its overall prediction. This paper tries to do a broad evaluation of uncertainty methods, and, particularly, to test how they perform on out of distribution data, including both data that is perturbed from its original values, and fully OOD data from ground-truth categories never seen during training. Ideally, we would want an uncertainty method that is less confident in its predictions as data is made more dissimilar from the distribution that the model is trained on. Some metrics the paper uses for capturing this are: 

- Brier Score (The difference between predicted score and ground truth 0/1 label, averaged over all examples)
- Negative Log Likelihood
- Expected Calibration Error (Within a given bucket, this is calculated as the difference between accuracy to ground truth labels, and the average predicted score in that bucket, capturing that you'd ideally want to have a lower predicted score in cases where you have low accuracy, and vice versa)
- Entropy - For labels that are fully out of distribution, and don't map to any of the model's categories, you can't directly calculate ground truth accuracy, but you can ideally ask for a model that has high entropy (close to uniform) probabilities over the classes it knows about when the image is drawn from an entirely different class

The authors test over image datasets small (MNIST) and large (ImageNet and CIFAR10), as well as a categorical ad-click-prediction dataset. They came up with some interesting findings. 

https://i.imgur.com/EVnjS1R.png

1. More fully principled Bayesian estimation of posteriors over parameters, in the form of Stochastic Variational Inference, works well on MNIST, but quite poorly on either categorical data or higher dimensional image datasets 

https://i.imgur.com/3emTYNP.png

2. Temperature scaling, which basically performs a second supervised calibration using a hold-out set to push your probabilities towards true probabilities, performs well in-distribution but collapses fairly quickly off-distribution (which sort of makes sense given that it too is just another supervised method that can do poorly when off-distribution) 
3. In general, ensemble methods, where you train different models on different subsets of the data and take their variance as uncertainty, perform the best across the bigger image models as well as the ad click model, likely because SVI (along with many other Bayesian methods) is too computationally intensive to get to work well on higher-dimensional data 
4. Overall, none of the methods worked particularly well, and even the best-performing ones were often confidently wrong off-distribution 

I think it's fair to say that we're far from where we wish we were when it comes to models that "know when they don't know," and this paper does a good job of highlighting that in specific fashion.