Recent years have witnessed the success of deep neural networks in dealing with a plenty of practical problems. Dropout has played an essential role in many successful deep neural networks, by inducing regularization in the model training. In this paper, we present a new regularized training approach: Shakeout. Instead of randomly discarding units as Dropout does at the training stage, Shakeout randomly chooses to enhance or reverse each unit's contribution to the next layer. This minor modification of Dropout has the statistical trait: the regularizer induced by Shakeout adaptively combines $L_0$, $L_1$ and $L_2$ regularization terms. Our classification experiments with representative deep architectures on image datasets MNIST, CIFAR-10 and ImageNet show that Shakeout deals with over-fitting effectively and outperforms Dropout. We empirically demonstrate that Shakeout leads to sparser weights under both unsupervised and supervised settings. Shakeout also leads to the grouping effect of the input units in a layer. Considering the weights in reflecting the importance of connections, Shakeout is superior to Dropout, which is valuable for the deep model compression. Moreover, we demonstrate that Shakeout can effectively reduce the instability of the training process of the deep architecture.

There has been appreciable progress in unsupervised network representation learning (UNRL) approaches over graphs recently with flexible random-walk approaches, new optimization objectives and deep architectures. However, there is no common ground for systematic comparison of embeddings to understand their behavior for different graphs and tasks. In this paper we theoretically group different approaches under a unifying framework and empirically investigate the effectiveness of different network representation methods. In particular, we argue that most of the UNRL approaches either explicitly or implicit model and exploit context information of a node. Consequently, we propose a framework that casts a variety of approaches -- random walk based, matrix factorization and deep learning based -- into a unified context-based optimization function. We systematically group the methods based on their similarities and differences. We study the differences among these methods in detail which we later use to explain their performance differences (on downstream tasks). We conduct a large-scale empirical study considering 9 popular and recent UNRL techniques and 11 real-world datasets with varying structural properties and two common tasks -- node classification and link prediction. We find that there is no single method that is a clear winner and that the choice of a suitable method is dictated by certain properties of the embedding methods, task and structural properties of the underlying graph. In addition we also report the common pitfalls in evaluation of UNRL methods and come up with suggestions for experimental design and interpretation of results.

LiveSketch is a novel algorithm for searching large image collections using hand-sketched queries. LiveSketch tackles the inherent ambiguity of sketch search by creating visual suggestions that augment the query as it is drawn, making query specification an iterative rather than one-shot process that helps disambiguate users' search intent. Our technical contributions are: a triplet convnet architecture that incorporates an RNN based variational autoencoder to search for images using vector (stroke-based) queries; real-time clustering to identify likely search intents (and so, targets within the search embedding); and the use of backpropagation from those targets to perturb the input stroke sequence, so suggesting alterations to the query in order to guide the search. We show improvements in accuracy and time-to-task over contemporary baselines using a 67M image corpus.

How can startups compete in Deep Learning when the tech giants (Google, Amazon, Baidu, Microsoft, Apple) have so much more data? There is this narrative out there that it is "all about the data," "whoever has the most data wins…" I may have subscribed to that theory in the past but I now think it is largely wrong, and may have been perpetuated by the big companies to tout their advantage. Deep Learning tools and frameworks are so nascent, and the skill set so rare, that meaningfully better algorithms are possible, and make a huge difference. For example, I believe Blue Hexagon has those. When there are literally 100's of thousands of new malware variants PER DAY, it just makes sense that IF you could get neural networks to analyze the traffic, at line speed, it would be order of magnitude better than the current signature and sandbox based approaches.

Artificial intelligence (AI) has become a critical aspect in financial services. Financial institutions around the world are making efforts to adopt AI for task automation, customer services, behavior analysis, as well as fraud finding, and are making large-scale investments in related technologies. The World Economic Forum (WEF) estimates the number to reach US$10 billion by 2020. In financial services, applications for AI technologies exist across nearly the entire spectrum of business, from algorithmic stock trading applications and credit card fraud detection, to auto investment advisors and market research and sentiment analysis. The following 10 AI fintech companies are some of Europe's rising stars to watch very closely: Swiss startup Parashift develops AI-based accounting document management technologies which it offers through a SaaS platform and APIs.

If you have observed, conventional audio and speech analysis systems are typically built using a pipeline structure, where the first step is to extract various low dimensional hand-crafted acoustic features (e.g., MFCC, pitch, RMSE, Chroma, and whatnot). Although hand-crafted acoustic features are typically well designed, is still not possible to retain all useful information due to the human knowledge bias and the high compression ratio. And of course, the feature engineering you will have to perform will depend on the type of audio problem that you are working on. But, how about learning directly from raw waveforms (i.e., raw audio files are directly fed into the deep neural network)? In this post, let's take learnings from this paper and try to apply it to the following Kaggle dataset. Go ahead and download the Heatbeat Sounds dataset.

The problem of explaining the behavior of deep neural networks has gained a lot of attention over the last years. While several attribution methods have been proposed, most come without strong theoretical foundations. This raises the question of whether the resulting attributions are reliable. On the other hand, the literature on cooperative game theory suggests Shapley values as a unique way of assigning relevance scores such that certain desirable properties are satisfied. Previous works on attribution methods also showed that explanations based on Shapley values better agree with the human intuition. Unfortunately, the exact evaluation of Shapley values is prohibitively expensive, exponential in the number of input features. In this work, by leveraging recent results on uncertainty propagation, we propose a novel, polynomial-time approximation of Shapley values in deep neural networks. We show that our method produces significantly better approximations of Shapley values than existing state-of-the-art attribution methods.

Recently, distant supervision has gained great success on Fine-grained Entity Typing (FET). Despite its efficiency in reducing manual labeling efforts, it also brings the challenge of dealing with false entity type labels, as distant supervision assigns labels in a context agnostic manner. Existing works alleviated this issue with partial-label loss, but usually suffer from confirmation bias, which means the classifier fit a pseudo data distribution given by itself. In this work, we propose to regularize distantly supervised models with Compact Latent Space Clustering (CLSC) to bypass this problem and effectively utilize noisy data yet. Our proposed method first dynamically constructs a similarity graph of different entity mentions; infer the labels of noisy instances via label propagation. Based on the inferred labels, mention embeddings are updated accordingly to encourage entity mentions with close semantics to form a compact cluster in the embedding space,thus leading to better classification performance. Extensive experiments on standard benchmarks show that our CLSC model consistently outperforms state-of-the-art distantly supervised entity typing systems by a significant margin.

We introduce a method to infer a variational approximation to the posterior distribution of solutions in computational imaging inverse problems. Machine learning methods applied to computational imaging have proven very successful, but have so far largely focused on retrieving a single optimal solution for a given task. Such retrieval is arguably an incomplete description of the solution space, as in ill-posed inverse problems there may be many similarly likely reconstructions. We minimise an upper bound on the divergence between our approximate distribution and the true intractable posterior, thereby obtaining a probabilistic description of the solution space in imaging inverse problems with empirical prior. We demonstrate the advantage of our technique in quantitative simulations with the CelebA dataset and common image reconstruction tasks. We then apply our method to two of the currently most challenging problems in experimental optics: imaging through highly scattering media and imaging through multi-modal optical fibres. In both settings we report state of the art reconstructions, while providing new capabilities, such as estimation of error-bars and visualisation of multiple likely reconstructions.

The prominence of deep learning, large amount of annotated data and increasingly powerful hardware made it possible to reach remarkable performance for supervised classification tasks, in many cases saturating the training sets. However, adapting the learned classification to new domains remains a hard problem due to at least three reasons: (1) the domains and the tasks might be drastically different; (2) there might be very limited amount of annotated data on the new domain and (3) full training of a new model for each new task is prohibitive in terms of memory, due to the shear number of parameter of deep networks. Instead, new tasks should be learned incrementally, building on prior knowledge from already learned tasks, and without catastrophic forgetting, i.e. without hurting performance on prior tasks. To our knowledge this paper presents the first method for multi-domain/task learning without catastrophic forgetting using a fully tensorized architecture. Our main contribution is a method for multi-domain learning which models groups of identically structured blocks within a CNN as a high-order tensor. We show that this joint modelling naturally leverages correlations across different layers and results in more compact representations for each new task/domain over previous methods which have focused on adapting each layer separately. We apply the proposed method to 10 datasets of the Visual Decathlon Challenge and show that our method offers on average about 7.5x reduction in number of parameters and superior performance in terms of both classification accuracy and Decathlon score. In particular, our method outperforms all prior work on the Visual Decathlon Challenge.