SGD is a widely used optimization method for training the parameters of some model f on some given task. Since the convergence of SGD is related to the variance of the stochastic gradient estimate, there's been a lot of work on trying to come up with such stochastic estimates with smaller variance. This paper does it using an importance sampling (IS) Monte Carlo estimate of the gradient, and learning the proposal distribution $q$ of the IS estimate. 

The proposal distribution $q$ is parametrized in some way, and is trained to minimize the variance of the gradient estimate. It is trained simultaneously while the model $f$ that SGD (i.e. the SGD that uses IS to get its gradient) is training. To make this whole story more recursive, the proposal distribution $q$ is also trained with SGD :-) This makes sense, since one expects the best proposal to depend on the value of the parameters of model $f$, so the best proposal $q$ should vary as $f$ is trained.

One application of this idea is in optimizing a classification model over a distribution that is imbalanced class-wise (e.g. there are classes with much fewer examples). In this case, the proposal distribution determines how frequently we sample examples from each class (conditioned on the class, training examples are chosen uniformly).


#### My two cents

This is a really cool idea. I particularly like the application to training on an imbalanced classification problem. People have mostly been using heuristics to tackle this problem, such as initially sampling each class equally as often, and then fine-tuning/calibrating the model using the real class proportions. This approach instead proposes a really elegant, coherent, solution to this problem.

I would have liked to see a comparison with that aforementioned heuristic (for mainly selfish reasons :-) ). They instead compare with an importance sampling approach with proposal that assigns the same probability to each class, which is a reasonable alternative (though I don't know if it's used as often as the more heuristic approach).

There are other applications, to matrix factorization and reinforcement learning, that are presented in the paper and seem neat, though I haven't gone through those as much.

Overall, one of my favorite paper this year: it's original, tackles a problem for which I've always hated the heuristic solution I'm using now, proposes an elegant solution to it, and is applicable even more widely than that setting.

This paper presents a recurrent neural network architecture in which some of the recurrent weights dynamically change during the forward pass, using a hebbian-like rule. They correspond to the matrices $A(t)$ in the figure below:

![Fast weights RNN figure](http://i.imgur.com/DCznSf4.png)

These weights $A(t)$ are referred to as *fast weights*. Comparatively, the recurrent weights $W$ are referred to as slow weights, since they are only changing due to normal training and are otherwise kept constant at test time.

More specifically, the proposed fast weights RNN compute a series of hidden states $h(t)$ over time steps $t$, but, unlike regular RNNs, the transition from $h(t)$ to $h(t+1)$ consists of multiple ($S$) recurrent layers $h_1(t+1), \dots, h_{S-1}(t+1), h_S(t+1)$, defined as follows:

$$h_{s+1}(t+1) = f(W h(t) + C x(t) + A(t) h_s(t+1))$$

where $f$ is an element-wise non-linearity such as the ReLU activation. The next hidden state $h(t+1)$ is simply defined as the last "inner loop" hidden state $h_S(t+1)$, before moving to the next time step. 

As for the fast weights $A(t)$, they too change between time steps, using the hebbian-like rule:

$$A(t+1) = \lambda A(t) + \eta h(t) h(t)^T$$

where $\lambda$ acts as a decay rate (to partially forget some of what's in the past)  and $\eta$ as the fast weight's "learning rate" (not to be confused with the learning rate used during backprop). Thus, the role played by the fast weights is to rapidly adjust to the recent hidden states and remember the recent past.

In fact, the authors show an explicit relation between these fast weights and memory-augmented architectures that have recently been popular. Indeed, by recursively applying and expending the equation for the fast weights, one obtains

$$A(t) = \eta \sum_{\tau = 1}^{\tau = t-1}\lambda^{t-\tau-1} h(\tau) h(\tau)^T$$

*(note the difference with Equation 3 of the paper... I think there was a typo)* which implies that when computing the $A(t) h_s(t+1)$ term in the expression to go from $h_s(t+1)$ to $h_{s+1}(t+1)$, this term actually corresponds to

$$A(t) h_s(t+1) = \eta \sum_{\tau =1}^{\tau = t-1} \lambda^{t-\tau-1} h(\tau) (h(\tau)^T h_s(t+1))$$

i.e. $A(t) h_s(t+1)$ is a weighted sum of all previous hidden states $h(\tau)$, with each hidden states weighted by an "attention weight" $h(\tau)^T h_s(t+1)$. The difference with many recent memory-augmented architectures is thus that the attention weights aren't computed using a softmax non-linearity.

Experimentally, they find it beneficial to use [layer normalization](https://arxiv.org/abs/1607.06450). Good values for $\eta$ and $\lambda$ seem to be 0.5 and 0.9 respectively. I'm not 100% sure, but I also understand that using $S=1$, i.e. using the fast weights only once per time steps, was usually found to be optimal. Also see Figure 3 for the architecture used on the image classification datasets, which is slightly more involved.

The authors present a series 4 experiments, comparing with regular RNNs (IRNNs, which are RNNs with ReLU units and whose recurrent weights are initialized to a scaled identity matrix) and LSTMs (as well as an associative LSTM for a synthetic associative retrieval task and ConvNets for the two image datasets). Generally, experiments illustrate that the fast weights RNN tends to train faster (in number of updates) and better than the other recurrent architectures. Surprisingly, the fast weights RNN can even be competitive with a ConvNet on the two image classification benchmarks, where the RNN traverses glimpses from the image using a fixed policy.

**My two cents**

This is a very thought provoking paper which, based on the comparison with LSTMs, suggests that fast weights RNNs might be a very good alternative. I'd be quite curious to see what would happen if one was to replace LSTMs with them in the myriad of papers using LSTMs (e.g. all the Seq2Seq work). Intuitively, LSTMs seem to be able to do more than just attending to the recent past. But, for a given task, if one was to observe that fast weights RNNs are competitive to LSTMs, it would suggests that the LSTM isn't doing something that much more complex. So it would be interesting to determine what are the tasks where the extra capacity of an LSTM is actually valuable and exploitable. Hopefully the authors will release some code, to facilitate this exploration. 

The discussion at the end of Section 3 on how exploiting the "memory augmented" view of fast weights is useful to allow the use of minibatches is interesting. However, it also suggests that computations in the fast weights RNN scales quadratically with the sequence size (since in this view, the RNN technically must attend to all previous hidden states, since the beginning of the sequence). This is something to keep in mind, if one was to consider applying this to very long sequences (i.e. much longer than the hidden state dimensionality).

Also, I don't quite get the argument that the "memory augmented" view of fast weights is more amenable to mini-batch training. I understand that having an explicit weight matrix $A(t)$ for each minibatch sequence complicates things. However, in the memory augmented view, we also have a "memory matrix" that is different for each sequence, and yet we can handle that fine. The problem I can imagine is that storing a *sequence of arbitrary weight matrices* for each sequence might be storage demanding (and thus perhaps make it impossible to store a forward/backward pass for more than one sequence at a time), while the implicit memory matrix only requires appending a new row at each time step. Perhaps the argument to be made here is more that there's already mini-batch compatible code out there for dealing with the use of a memory matrix of stored previous memory states.

This work strikes some (partial) resemblance to other recent work, which may serve as food for thought here. The use of possibly multiple computation layers between time steps reminds me of [Adaptive Computation Time (ACT) RNN]( http://www.shortscience.org/paper?bibtexKey=journals/corr/Graves16). Also, expressing a backpropable architecture that involves updates to weights (here, hebbian-like updates) reminds me of recent work that does backprop through the updates of a gradient descent procedure (for instance as in [this work]( http://www.shortscience.org/paper?bibtexKey=conf/icml/MaclaurinDA15)). 

Finally, while I was familiar with the notion of fast weights from the work on [Using Fast Weights to Improve Persistent Contrastive Divergence](http://people.ee.duke.edu/~lcarin/FastGibbsMixing.pdf), I didn't realize that this concept dated as far back as the late 80s. So, for young researchers out there looking for inspiration for research ideas, this paper confirms that looking at the older neural network literature for inspiration is probably a very good strategy :-)

To sum up, this is really nice work, and I'm looking forward to the NIPS 2016 oral presentation of it!

This paper from 2016 introduced a new k-mer based method to estimate isoform abundance from RNA-Seq data called kallisto.  The method provided a significant improvement in speed and memory usage compared to the previously used methods while yielding similar accuracy.   In fact, kallisto is able to quantify expression in a matter of minutes instead of hours.

The standard (previous) methods for quantifying expression rely on mapping, i.e. on the alignment of a transcriptome sequenced reads to a genome of reference.  Reads are assigned to a position in the genome and the gene or isoform expression values are derived by counting the number of reads overlapping the features of interest. 

The idea behind kallisto is to rely on a pseudoalignment which does not attempt to identify the positions of the reads in the transcripts, only the potential transcripts of origin. Thus,  it avoids doing an alignment of each read to a reference genome. In fact, kallisto only uses the transcriptome sequences (not the whole genome) in its first step which is the generation of  the kallisto index.  Kallisto builds a colored de Bruijn graph (T-DBG) from all the k-mers found in the transcriptome.  Each node of the graph corresponds to a k-mer (a short sequence of k nucleotides) and retains the information about the transcripts in which they can be found in the form of a color.  Linear stretches having the same coloring in the graph correspond to transcripts. Once the T-DBG is built, kallisto stores a hash table mapping each k-mer to its transcript(s) of origin along with the position within the transcript(s).  This step is done only once and is dependent on a provided annotation file (containing the sequences of all the transcripts in the transcriptome).  
  
Then for a given sequenced sample, kallisto decomposes each read into its k-mers and uses those k-mers to find a path covering in the T-DBG.  This path covering of the transcriptome graph, where a path corresponds to a transcript, generates k-compatibility classes for each k-mer, i.e. sets of potential transcripts of origin on the nodes.   The potential transcripts of origin for a read can be obtained using the intersection of its k-mers k-compatibility classes. To make the pseudoalignment faster, kallisto removes redundant k-mers since neighboring k-mers often belong to the same transcripts. Figure1, from the paper, summarizes these different steps.

https://i.imgur.com/eNH2kuO.png

**Figure1**. Overview of kallisto. The input consists of a reference transcriptome and reads from an RNA-seq experiment. (a) An example of a read (in black) and three overlapping transcripts with exonic regions as shown. (b) An index is constructed by creating the transcriptome de Bruijn Graph (T-DBG) where nodes (v1, v2, v3, ... ) are k-mers, each transcript corresponds to a colored path as shown and the path cover of the transcriptome induces a k-compatibility class for each k-mer. (c) Conceptually, the k-mers of a read are hashed (black nodes) to find the k-compatibility class of a read. (d) Skipping (black dashed lines) uses the information stored in the T-DBG to skip k-mers that are redundant because they have the same k-compatibility class. (e) The k-compatibility class of the read is determined by taking the intersection of the k-compatibility classes of its constituent k-mers.[From Bray et al. Near-optimal probabilistic RNA-seq quantification, Nature Biotechnology, 2016.]

Then, kallisto optimizes the following RNA-Seq likelihood function using the expectation-maximization (EM) algorithm.  

$$L(\alpha) \propto \prod_{f \in F} \sum_{t \in T} y_{f,t} \frac{\alpha_t}{l_t} = \prod_{e \in E}\left(  \sum_{t \in e} \frac{\alpha_t}{l_t} \right )^{c_e}$$

In this function,  $F$ is the set of fragments (or reads), $T$ is the set of transcripts, $l_t$ is the (effective) length of transcript $t$ and **y**$_{f,t}$ is a compatibility matrix defined as 1 if  fragment $f$ is compatible with $t$ and 0 otherwise.  The parameters $α_t$ are the probabilities of selecting reads from a transcript $t$.  These $α_t$ are the parameters of interest since they represent the isoforms abundances or relative expressions.

To make things faster, the compatibility matrix is collapsed (factorized) into equivalence classes. An equivalent class consists of all the reads compatible with the same subsets of transcripts. The EM algorithm is applied to equivalence classes (not to reads).  Each $α_t$ will be optimized to maximise the likelihood of transcript abundances given observations of the equivalence classes. The speed of the method makes it possible to evaluate the uncertainty of the  abundance estimates for each RNA-Seq sample using a bootstrap technique.  For a given sample containing $N$ reads, a bootstrap sample is generated from the sampling of $N$ counts from a multinomial distribution over the equivalence classes derived from the original sample.  The EM algorithm is applied on those sampled equivalence class counts to estimate transcript abundances. The bootstrap information is then used in downstream analyses such as determining which genes are differentially expressed.

Practically, we can illustrate the different steps involved in kallisto using a small example.  Starting from a tiny genome with 3 transcripts, assume that the RNA-Seq experiment produced 4 reads as depicted in the image below.

https://i.imgur.com/5JDpQO8.png

The first step is to build the T-DBG graph and the kallisto index.  All transcript sequences are decomposed into k-mers (here k=5) to construct the colored de Bruijn graph. Not all nodes are represented in the following drawing.  The idea is that each different transcript will lead to a different path in the graph.  The strand is not taken into account, kallisto is strand-agnostic.

https://i.imgur.com/4oW72z0.png

Once the index is built, the four reads of the sequenced sample can be analysed.  They are decomposed into k-mers (k=5 here too) and the pre-built index is used to determine the k-compatibility class of each k-mer. Then, the k-compatibility class of each read is computed. For example, for read 1, the intersection of all the k-compatibility classes of its k-mers suggests that it might come from transcript 1 or transcript 2.

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

This is done for the four reads enabling the construction of the compatibility matrix  **y**$_{f,t}$ which is part of the RNA-Seq likelihood function.  In this equation, the $α_t$ are the parameters that we want to estimate.

https://i.imgur.com/Hp5QJvH.png

The EM algorithm being too slow to be applied on millions of reads, the compatibility matrix **y**$_{f,t}$ is factorized into equivalence classes and a count is computed for each class (how many reads are represented by this equivalence class). The EM algorithm uses this collapsed information to maximize the new equivalent RNA-Seq likelihood function and optimize the $α_t$.

https://i.imgur.com/qzsEq8A.png

The EM algorithm stops when for every transcript $t$, $α_tN$ > 0.01 changes less than 1%, where $N$ is the total number of reads. 

**Goal**: identifying training points most responsible for a given prediction.

Given training points $z_1, \dots, z_n$, let loss function be $\frac{1}{n}\sum_{i=1}^nL(z_i, \theta)$ 

A function called influence function let us compute the parameter change if $z$ were upweighted by some small $\epsilon$. 
$$\hat{\theta}_{\epsilon, z} := \arg \min_{\theta \in \Theta} \frac{1}{n}\sum_{i=1}^n L(z_i, \theta) + \epsilon L(z, \theta)$$

$$\mathcal{I}_{\text{up, params}}(z) := \frac{d\hat{\theta}_{\epsilon, z}}{d\epsilon} = -H_{\hat{\theta}}^{-1} \nabla_\theta L(z, \hat{\theta})$$

$\mathcal{I}_{\text{up, params}}(z)$ shows how uplifting one point $z$ affect the estimate of the parameters $\theta$. 

Furthermore, we could determine how uplifting $z$ affect the loss estimate of a test point through chain rule. 
$$\mathcal{I}_{\text{up, loss}}(z, z_{\text{test}}) = \nabla_\theta L(z_{\text{test}}, \hat{\theta})^\top \mathcal{I}_{\text{up, params}}(z)$$ 

Apart from lifting one training point, change of the parameters with the change of a training point could also be estimated. 
$$\frac{d\hat{\theta}_{\epsilon, z_\delta, -z}}{d\epsilon} = \mathcal{I}_{\text{up, params}}(z_\delta) - \mathcal{I}_{\text{up, params}}(z)$$
This measures how purturbation $\delta$ to training point $z$ affect the parameter estimation $\theta$.

Section 3 describes some practicals about efficient implementing.

This set of tool could be used for some interpretable machine learning tasks.

This is follow-up work to the ResNets paper. It studies the propagation formulations behind the connections of deep residual networks and performs ablation experiments. A residual block can be represented with the equations $y_l = h(x_l) + F(x_l, W_l); x_{l+1} = f(y_l)$. $x_l$ is the input to the l-th unit and $x_{l+1}$ is the output of the l-th unit. In the original ResNets paper, $h(x_l) = x_l$, $f$ is ReLu, and F consists of 2-3 convolutional layers (bottleneck architecture) with BN and ReLU in between. In this paper, they propose a residual block with both $h(x)$ and $f(x)$ as identity mappings, which trains faster and performs better than their earlier baseline. Main contributions:

- Identity skip connections work much better than other multiplicative interactions that they experiment with:
    - Scaling $(h(x) = \lambda x)$: Gradients can explode or vanish depending on whether modulating scalar \lambda > 1 or < 1.
    - Gating ($1-g(x)$ for skip connection and $g(x)$ for function F):
    For gradients to propagate freely, $g(x)$ should approach 1, but
    F gets suppressed, hence suboptimal. This is similar to highway
    networks. $g(x)$ is a 1x1 convolutional layer.
    - Gating (shortcut-only): Setting high biases pushes initial $g(x)$
    towards identity mapping, and test error is much closer to baseline.
    - 1x1 convolutional shortcut: These work well for shallower networks
    (~34 layers), but training error becomes high for deeper networks,
    probably because they impede gradient propagation.

- Experiments on activations.
    - BN after addition messes up information flow, and performs considerably
    worse.
    - ReLU before addition forces the signal to be non-negative, so the signal is monotonically increasing, while ideally a residual function should be free to take values in (-inf, inf).
    - BN + ReLU pre-activation works best. This also prevents overfitting, due
    to BN's regularizing effect. Input signals to all weight layers are normalized.

## Strengths

- Thorough set of experiments to show that identity shortcut connections
are easiest for the network to learn. Activation of any deeper unit can
be written as the sum of the activation of a shallower unit and a residual
function. This also implies that gradients can be directly propagated to
shallower units. This is in contrast to usual feedforward networks, where
gradients are essentially a series of matrix-vector products, that may vanish, as networks grow deeper.

- Improved accuracies than their previous ResNets paper.

## Weaknesses / Notes

- Residual units are useful and share the same core idea that worked in
LSTM units. Even though stacked non-linear layers are capable of asymptotically
approximating any arbitrary function, it is clear from recent work that
residual functions are much easier to approximate than the complete function.
The [latest Inception paper](http://arxiv.org/abs/1602.07261) also reports
that training is accelerated and performance is improved by using identity
skip connections across Inception modules.

- It seems like the degradation problem, which serves as motivation for
residual units, exists in the first place for non-idempotent activation
functions such as sigmoid, hyperbolic tan. This merits further
investigation, especially with recent work on function-preserving transformations such as [Network Morphism](http://arxiv.org/abs/1603.01670), which expands the Net2Net idea to sigmoid, tanh, by using parameterized activations, initialized to identity mappings.