The composition of polyphonic chorale music in the style of J.S Bach has represented a major challenge in automatic music composition over the last decades. The art of Bach chorales composition involves combining four-part harmony with characteristic rhythmic patterns and typical melodic movements to produce musical phrases which begin, evolve and end (cadences) in a harmonious way. To our knowledge, no model so far was able to solve all these problems simultaneously using an agnostic machine-learning approach. This paper introduces DeepBach, a statistical model aimed at modeling polyphonic music and specifically four parts, hymn-like pieces. We claim that, after being trained on the chorale harmonizations by Johann Sebastian Bach, our model is capable of generating highly convincing chorales in the style of Bach. We evaluate how indistinguishable our generated chorales are from existing Bach chorales with a listening test. The results corroborate our claim. A key strength of DeepBach is that it is agnostic and flexible. Users can constrain the generation by imposing some notes, rhythms or cadences in the generated score. This allows users to reharmonize user-defined melodies. DeepBach's generation is fast, making it usable for interactive music composition applications. Several generation examples are provided and discussed from a musical point of view.
Captured tweets and retweets: 44
Problems at the intersection of vision and language are of significant importance both as challenging research questions and for the rich set of applications they enable. However, inherent structure in our world and bias in our language tend to be a simpler signal for learning than visual modalities, resulting in models that ignore visual information, leading to an inflated sense of their capability. We propose to counter these language priors for the task of Visual Question Answering (VQA) and make vision (the V in VQA) matter! Specifically, we balance the popular VQA dataset (Antol et al., ICCV 2015) by collecting complementary images such that every question in our balanced dataset is associated with not just a single image, but rather a pair of similar images that result in two different answers to the question. Our dataset is by construction more balanced than the original VQA dataset and has approximately twice the number of image-question pairs. Our complete balanced dataset will be publicly released as part of the 2nd iteration of the Visual Question Answering Challenge (VQA v2.0). We further benchmark a number of state-of-art VQA models on our balanced dataset. All models perform significantly worse on our balanced dataset, suggesting that these models have indeed learned to exploit language priors. This finding provides the first concrete empirical evidence for what seems to be a qualitative sense among practitioners. Finally, our data collection protocol for identifying complementary images enables us to develop a novel interpretable model, which in addition to providing an answer to the given (image, question) pair also provides a counter-example based explanation - specifically, it identifies an image that is similar to the original image, but it believes has a different answer to the same question. This can help in building trust for machines among their users.
Captured tweets and retweets: 2
The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this limitation and train networks that can maintain expertise on tasks which they have not experienced for a long time. Our approach remembers old tasks by selectively slowing down learning on the weights important for those tasks. We demonstrate our approach is scalable and effective by solving a set of classification tasks based on the MNIST hand written digit dataset and by learning several Atari 2600 games sequentially.
Captured tweets and retweets: 1
Automatic image synthesis research has been rapidly growing with deep networks getting more and more expressive. In the last couple of years, we have observed images of digits, indoor scenes, birds, chairs, etc. being automatically generated. The expressive power of image generators have also been enhanced by introducing several forms of conditioning variables such as object names, sentences, bounding box and key-point locations. In this work, we propose a novel deep conditional generative adversarial network architecture that takes its strength from the semantic layout and scene attributes integrated as conditioning variables. We show that our architecture is able to generate realistic outdoor scene images under different conditions, e.g. day-night, sunny-foggy, with clear object boundaries.
Captured tweets and retweets: 23
Recently, various deep-neural-network (DNN)-based approaches have been proposed for single-image super-resolution (SISR). Despite their promising results on major structure regions such as edges and lines, they still suffer from limited performance on texture regions that consist of very complex and fine patterns. This is because, during the acquisition of a low-resolution (LR) image via down-sampling, these regions lose most of the high frequency information necessary to represent the texture details. In this paper, we present a novel texture enhancement framework for SISR to effectively improve the spatial resolution in the texture regions as well as edges and lines. We call our method, high-resolution (HR) style transfer algorithm. Our framework consists of three steps: (i) generate an initial HR image from an interpolated LR image via an SISR algorithm, (ii) generate an HR style image from the initial HR image via down-scaling and tiling, and (iii) combine the HR style image with the initial HR image via a customized style transfer algorithm. Here, the HR style image is obtained by down-scaling the initial HR image and then repetitively tiling it into an image of the same size as the HR image. This down-scaling and tiling process comes from the idea that texture regions are often composed of small regions that similar in appearance albeit sometimes different in scale. This process creates an HR style image that is rich in details, which can be used to restore high-frequency texture details back into the initial HR image via the style transfer algorithm. Experimental results on a number of texture datasets show that our proposed HR style transfer algorithm provides more visually pleasing results compared with competitive methods.
Captured tweets and retweets: 25
Generating high-resolution, photo-realistic images has been a long-standing goal in machine learning. Recently, Nguyen et al. (2016) showed one interesting way to synthesize novel images by performing gradient ascent in the latent space of a generator network to maximize the activations of one or multiple neurons in a separate classifier network. In this paper we extend this method by introducing an additional prior on the latent code, improving both sample quality and sample diversity, leading to a state-of-the-art generative model that produces high quality images at higher resolutions (227x227) than previous generative models, and does so for all 1000 ImageNet categories. In addition, we provide a unified probabilistic interpretation of related activation maximization methods and call the general class of models "Plug and Play Generative Networks". PPGNs are composed of 1) a generator network G that is capable of drawing a wide range of image types and 2) a replaceable "condition" network C that tells the generator what to draw. We demonstrate the generation of images conditioned on a class (when C is an ImageNet or MIT Places classification network) and also conditioned on a caption (when C is an image captioning network). Our method also improves the state of the art of Multifaceted Feature Visualization, which generates the set of synthetic inputs that activate a neuron in order to better understand how deep neural networks operate. Finally, we show that our model performs reasonably well at the task of image inpainting. While image models are used in this paper, the approach is modality-agnostic and can be applied to many types of data.
Captured tweets and retweets: 2
Recent advances in deep learning have shown exciting promise in filling large holes in natural images with semantically plausible and context aware details, impacting fundamental image manipulation tasks such as object removal. While these learning-based methods are significantly more effective in capturing high-level features than prior techniques, they can only handle very low-resolution inputs due to memory limitations and difficulty in training. Even for slightly larger images, the inpainted regions would appear blurry and unpleasant boundaries become visible. We propose a multi-scale neural patch synthesis approach based on joint optimization of image content and texture constraints, which not only preserves contextual structures but also produces high-frequency details by matching and adapting patches with the most similar mid-layer feature correlations of a deep classification network. We evaluate our method on the ImageNet and Paris Streetview datasets and achieved state-of-the-art inpainting accuracy. We show our approach produces sharper and more coherent results than prior methods, especially for high-resolution images.
Captured tweets and retweets: 32
High-level manipulation of facial expressions in images --- such as changing a smile to a neutral expression --- is challenging because facial expression changes are highly non-linear, and vary depending on the appearance of the face. We present a fully automatic approach to editing faces that combines the advantages of flow-based face manipulation with the more recent generative capabilities of Variational Autoencoders (VAEs). During training, our model learns to encode the flow from one expression to another over a low-dimensional latent space. At test time, expression editing can be done simply using latent vector arithmetic. We evaluate our methods on two applications: 1) single-image facial expression editing, and 2) facial expression interpolation between two images. We demonstrate that our method generates images of higher perceptual quality than previous VAE and flow-based methods.
Captured tweets and retweets: 16
This paper presents a framework to tackle combinatorial optimization problems using neural networks and reinforcement learning. We focus on the traveling salesman problem (TSP) and train a recurrent network that, given a set of city coordinates, predicts a distribution over different city permutations. Using negative tour length as the reward signal, we optimize the parameters of the recurrent network using a policy gradient method. We compare learning the network parameters on a set of training graphs against learning them on individual test graphs. The best results are obtained when the network is first optimized on a training set and then refined on individual test graphs. Without any supervision and with minimal engineering, Neural Combinatorial Optimization achieves close to optimal results on 2D Euclidean graphs with up to 100 nodes.
Captured tweets and retweets: 2
We consider the problem of face swapping in images, where an input identity is transformed into a target identity while preserving pose, facial expression, and lighting. To perform this mapping, we use convolutional neural networks trained to capture the appearance of the target identity from an unstructured collection of his/her photographs.This approach is enabled by framing the face swapping problem in terms of style transfer, where the goal is to render an image in the style of another one. Building on recent advances in this area, we devise a new loss function that enables the network to produce highly photorealistic results. By combining neural networks with simple pre- and post-processing steps, we aim at making face swap work in real-time with no input from the user.
Captured tweets and retweets: 72
In this work, we explore the problem of generating fantastic special-effects for the typography. It is quite challenging due to the model diversities to illustrate varied text effects for different characters. To address this issue, our key idea is to exploit the analytics on the high regularity of the spatial distribution for text effects to guide the synthesis process. Specifically, we characterize the stylized patches by their normalized positions and the optimal scales to depict their style elements. Our method first estimates these two features and derives their correlation statistically. They are then converted into soft constraints for texture transfer to accomplish adaptive multi-scale texture synthesis and to make style element distribution uniform. It allows our algorithm to produce artistic typography that fits for both local texture patterns and the global spatial distribution in the example. Experimental results demonstrate the superiority of our method for various text effects over conventional style transfer methods. In addition, we validate the effectiveness of our algorithm with extensive artistic typography library generation.
Captured tweets and retweets: 57
Prepositions are very common and very ambiguous, and understanding their sense is critical for understanding the meaning of the sentence. Supervised corpora for the preposition-sense disambiguation task are small, suggesting a semi-supervised approach to the task. We show that signals from unannotated multilingual data can be used to improve supervised preposition-sense disambiguation. Our approach pre-trains an LSTM encoder for predicting the translation of a preposition, and then incorporates the pre-trained encoder as a component in a supervised classification system, and fine-tunes it for the task. The multilingual signals consistently improve results on two preposition-sense datasets.
Captured tweets and retweets: 2
We introduce the task of Visual Dialog, which requires an AI agent to hold a meaningful dialog with humans in natural, conversational language about visual content. Specifically, given an image, a dialog history, and a question about the image, the agent has to ground the question in image, infer context from history, and answer the question accurately. Visual Dialog is disentangled enough from a specific downstream task so as to serve as a general test of machine intelligence, while being grounded in vision enough to allow objective evaluation of individual responses and benchmark progress. We develop a novel two-person chat data-collection protocol to curate a large-scale Visual Dialog dataset (VisDial). Data collection is underway and on completion, VisDial will contain 1 dialog with 10 question-answer pairs on all ~200k images from COCO, with a total of 2M dialog question-answer pairs. We introduce a family of neural encoder-decoder models for Visual Dialog with 3 encoders -- Late Fusion, Hierarchical Recurrent Encoder and Memory Network -- and 2 decoders (generative and discriminative), which outperform a number of sophisticated baselines. We propose a retrieval-based evaluation protocol for Visual Dialog where the AI agent is asked to sort a set of candidate answers and evaluated on metrics such as mean-reciprocal-rank of human response. We quantify gap between machine and human performance on the Visual Dialog task via human studies. Our dataset, code, and trained models will be released publicly. Putting it all together, we demonstrate the first 'visual chatbot'!
Captured tweets and retweets: 2
Objective: Brain-Computer Interface technologies (BCI) enable the direct communication between humans and computers by analyzing brain measurements, such as electroencephalography (EEG). These technologies have been applied to a variety of domains, including neuroprosthetic control and the monitoring of epileptic seizures. Existing BCI systems primarily use a priori knowledge of EEG features of interest to build machine learning models. Recently, convolutional networks have been used for automatic feature extraction of large image databases, where they have obtained state-of-the-art results. In this work we introduce EEGNet, a compact fully convolutional network for EEG-based BCIs developed using Deep Learning approaches. Methods: EEGNet is a 4-layer convolutional network that uses filter factorization for learning a compact representation of EEG time series. EEGNet is one of the smallest convolutional networks to date, having less than 2200 parameters for a binary classification. Results: We show state-of-the-art classification performance across four different BCI paradigms: P300 event-related potential, error-related negativity, movement-related cortical potential, and sensory motor rhythm, with as few as 500 EEG trials. We also show that adding more trials reduces the error variance of prediction rather than improving classification performance. Conclusion: We provide preliminary evidence suggesting that our model can be used with small EEG databases while improving upon the state-of-the-art performance across several tasks and across subjects. Significance: The EEGNet neural network architecture provides state-of-the-art performance across several tasks and across subjects, challenging the notion that large datasets are required to obtain optimal performance.
Captured tweets and retweets: 2
We introduce GuessWhat?!, a two-player guessing game as a testbed for research on the interplay of computer vision and dialogue systems. The goal of the game is to locate an unknown object in a rich image scene by asking a sequence of questions. Higher-level image understanding, like spatial reasoning and language grounding, is required to solve the proposed task. Our key contribution is the collection of a large-scale dataset consisting of 150K human-played games with a total of 800K visual question-answer pairs on 66K images. We explain our design decisions in collecting the dataset and introduce the oracle and questioner tasks that are associated with the two players of the game. We prototyped deep learning models to establish initial baselines of the introduced tasks.
Captured tweets and retweets: 2
Neural Style Transfer has shown very exciting results enabling new forms of image manipulation. Here we extend the existing method beyond the paradigm of transferring global style information between pairs of images. In particular, we introduce control over spatial location, colour information and across spatial scale. We demonstrate how this enhances the method by allowing high-resolution controlled stylisation and helps to alleviate common failure cases such as applying ground textures to sky regions. Furthermore, by decomposing style into these perceptual factors we enable the combination of style information from multiple sources to generate new, perceptually appealing styles from existing ones. Finally we show how the introduced control measures can be applied in recent methods for Fast Neural Style Transfer.
Captured tweets and retweets: 29
We investigate conditional adversarial networks as a general-purpose solution to image-to-image translation problems. These networks not only learn the mapping from input image to output image, but also learn a loss function to train this mapping. This makes it possible to apply the same generic approach to problems that traditionally would require very different loss formulations. We demonstrate that this approach is effective at synthesizing photos from label maps, reconstructing objects from edge maps, and colorizing images, among other tasks. As a community, we no longer hand-engineer our mapping functions, and this work suggests we can achieve reasonable results without hand-engineering our loss functions either.
Captured tweets and retweets: 127
Speech is one of the most effective ways of communication among humans. Even though audio is the most common way of transmitting speech, very important information can be found in other modalities, such as vision. Vision is particularly useful when the acoustic signal is corrupted. Multi-modal speech recognition however has not yet found wide-spread use, mostly because the temporal alignment and fusion of the different information sources is challenging. This paper presents an end-to-end audiovisual speech recognizer (AVSR), based on recurrent neural networks (RNN) with a connectionist temporal classification (CTC) loss function. CTC creates sparse "peaky" output activations, and we analyze the differences in the alignments of output targets (phonemes or visemes) between audio-only, video-only, and audio-visual feature representations. We present the first such experiments on the large vocabulary IBM ViaVoice database, which outperform previously published approaches on phone accuracy in clean and noisy conditions.
Captured tweets and retweets: 7
This work brings together two powerful concepts in Gaussian processes: the variational approach to sparse approximation and the spectral representation of Gaussian processes. This gives rise to an approximation that inherits the benefits of the variational approach but with the representational power and computational scalability of spectral representations. The work hinges on a key result that there exist spectral features related to a finite domain of the Gaussian process which exhibit almost-independent covariances. We derive these expressions for Matern kernels in one dimension, and generalize to more dimensions using kernels with specific structures. Under the assumption of additive Gaussian noise, our method requires only a single pass through the dataset, making for very fast and accurate computation. We fit a model to 4 million training points in just a few minutes on a standard laptop. With non-conjugate likelihoods, our MCMC scheme reduces the cost of computation from O(NM2) (for a sparse Gaussian process) to O(NM) per iteration, where N is the number of data and M is the number of features.
Captured tweets and retweets: 4
Generative Adversarial Networks (GANs) have recently demonstrated to successfully approximate complex data distributions. A relevant extension of this model is conditional GANs (cGANs), where the introduction of external information allows to determine specific representations of the generated images. In this work, we evaluate encoders to inverse the mapping of a cGAN, i.e., mapping a real image into a latent space and a conditional representation. This allows, for example, to reconstruct and modify real images of faces conditioning on arbitrary attributes. Additionally, we evaluate the design of cGANs. The combination of an encoder with a cGAN, which we call Invertible cGAN (IcGAN), enables to re-generate real images with deterministic complex modifications.
Captured tweets and retweets: 15