The goal of a generative model is to capture the distribution underlying the data, typically through latent variables. After training, these variables are often used as a new representation, more effective than the original features in a variety of learning tasks. However, the representations constructed by contemporary generative models are usually point-wise deterministic mappings from the original feature space. Thus, even with representations robust to class-specific transformations, statistically driven models trained on them would not be able to generalize when the labeled data is scarce. Inspired by the stochasticity of the synaptic connections in the brain, we introduce Energy-based Stochastic Ensembles. These ensembles can learn non-deterministic representations, i.e., mappings from the feature space to a family of distributions in the latent space. These mappings are encoded in a distribution over a (possibly infinite) collection of models. By conditionally sampling models from the ensemble, we obtain multiple representations for every input example and effectively augment the data. We propose an algorithm similar to contrastive divergence for training restricted Boltzmann stochastic ensembles. Finally, we demonstrate the concept of the stochastic representations on a synthetic dataset as well as test them in the one-shot learning scenario on MNIST.
Traditional techniques for measuring similarities between time series are based on handcrafted similarity measures, whereas more recent learning-based approaches cannot exploit external supervision. We combine ideas from time-series modeling and metric learning, and study siamese recurrent networks (SRNs) that minimize a classification loss to learn a good similarity measure between time series. Specifically, our approach learns a vectorial representation for each time series in such a way that similar time series are modeled by similar representations, and dissimilar time series by dissimilar representations. Because it is a similarity prediction models, SRNs are particularly well-suited to challenging scenarios such as signature recognition, in which each person is a separate class and very few examples per class are available. We demonstrate the potential merits of SRNs in within-domain and out-of-domain classification experiments and in one-shot learning experiments on tasks such as signature, voice, and sign language recognition.
Despite recent breakthroughs in the applications of deep neural networks, one setting that presents a persistent challenge is that of "one-shot learning." Traditional gradient-based networks require a lot of data to learn, often through extensive iterative training. When new data is encountered, the models must inefficiently relearn their parameters to adequately incorporate the new information without catastrophic interference. Architectures with augmented memory capacities, such as Neural Turing Machines (NTMs), offer the ability to quickly encode and retrieve new information, and hence can potentially obviate the downsides of conventional models. Here, we demonstrate the ability of a memory-augmented neural network to rapidly assimilate new data, and leverage this data to make accurate predictions after only a few samples. We also introduce a new method for accessing an external memory that focuses on memory content, unlike previous methods that additionally use memory location-based focusing mechanisms.
In this paper, we introduce an end-to-end framework for video analysis focused towards practical scenarios built on theoretical foundations from sparse representation, including a novel descriptor for general purpose video analysis. In our approach, we compute kinematic features from optical flow and first and second-order derivatives of intensities to represent motion and appearance respectively. These features are then used to construct covariance matrices which capture joint statistics of both low-level motion and appearance features extracted from a video. Using an over-complete dictionary of the covariance based descriptors built from labeled training samples, we formulate low-level event recognition as a sparse linear approximation problem. Within this, we pose the sparse decomposition of a covariance matrix, which also conforms to the space of semi-positive definite matrices, as a determinant maximization problem. Also since covariance matrices lie on non-linear Riemannian manifolds, we compare our former approach with a sparse linear approximation alternative that is suitable for equivalent vector spaces of covariance matrices. This is done by searching for the best projection of the query data on a dictionary using an Orthogonal Matching pursuit algorithm. We show the applicability of our video descriptor in two different application domains - namely low-level event recognition in unconstrained scenarios and gesture recognition using one shot learning. Our experiments provide promising insights in large scale video analysis.
One-shot learning is usually tackled by using generative models or discriminative embeddings. Discriminative methods based on deep learning, which are very effective in other learning scenarios, are ill-suited for one-shot learning as they need large amounts of training data. In this paper, we propose a method to learn the parameters of a deep model in one shot. We construct the learner as a second deep network, called a learnet, which predicts the parameters of a pupil network from a single exemplar. In this manner we obtain an efficient feed-forward one-shot learner, trained end-to-end by minimizing a one-shot classification objective in a learning to learn formulation. In order to make the construction feasible, we propose a number of factorizations of the parameters of the pupil network. We demonstrate encouraging results by learning characters from single exemplars in Omniglot, and by tracking visual objects from a single initial exemplar in the Visual Object Tracking benchmark.
One of the key challenges in applying reinforcement learning to complex robotic control tasks is the need to gather large amounts of experience in order to find an effective policy for the task at hand. Model-based reinforcement learning can achieve good sample efficiency, but requires the ability to learn a model of the dynamics that is good enough to learn an effective policy. In this work, we develop a model-based reinforcement learning algorithm that combines prior knowledge from previous tasks with online adaptation of the dynamics model. These two ingredients enable highly sample-efficient learning even in regimes where estimating the true dynamics is very difficult, since the online model adaptation allows the method to locally compensate for unmodeled variation in the dynamics. We encode the prior experience into a neural network dynamics model, adapt it online by progressively refitting a local linear model of the dynamics, and use model predictive control to plan under these dynamics. Our experimental results show that this approach can be used to solve a variety of complex robotic manipulation tasks in just a single attempt, using prior data from other manipulation behaviors.
The cost function used to train a generative model should fit the purpose of the model. If the model is intended for tasks such as generating perceptually correct samples, it is beneficial to maximise the likelihood of a sample drawn from the model, Q, coming from the same distribution as the training data, P. This is equivalent to minimising the Kullback-Leibler (KL) distance, KL[Q||P]. However, if the model is intended for tasks such as retrieval or classification it is beneficial to maximise the likelihood that a sample drawn from the training data is captured by the model, equivalent to minimising KL[P||Q]. The cost function used in adversarial training optimises the Jensen-Shannon entropy which can be seen as an even interpolation between KL[Q||P] and KL[P||Q]. Here, we propose an alternative adversarial cost function which allows easy tuning of the model for either task. Our task specific cost function is evaluated on a dataset of hand-written characters in the following tasks: Generation, retrieval and one-shot learning.
Hebbian plasticity is a powerful principle that allows biological brains to learn from their lifetime experience. By contrast, artificial neural networks trained with backpropagation generally have fixed connection weights that do not change once training is complete. While recent methods can endow neural networks with long-term memories, Hebbian plasticity is currently not amenable to gradient descent. Here we derive analytical expressions for activity gradients in neural networks with Hebbian plastic connections. Using these expressions, we can use backpropagation to train not just the baseline weights of the connections, but also their plasticity. As a result, the networks "learn how to learn" in order to solve the problem at hand: the trained networks automatically perform fast learning of unpredictable environmental features during their lifetime, expanding the range of solvable problems. We test the algorithm on various on-line learning tasks, including pattern completion, one-shot learning, and reversal learning. The algorithm successfully learns how to learn the relevant associations from one-shot instruction, and fine-tunes the temporal dynamics of plasticity to allow for continual learning in response to changing environmental parameters. We conclude that backpropagation of Hebbian plasticity offers a powerful model for lifelong learning.
Automatic translation from natural language descriptions into programs is a longstanding challenging problem. In this work, we consider a simple yet important sub-problem: translation from textual descriptions to If-Then programs. We devise a novel neural network architecture for this task which we train end-to-end. Specifically, we introduce Latent Attention, which computes multiplicative weights for the words in the description in a two-stage process with the goal of better leveraging the natural language structures that indicate the relevant parts for predicting program elements. Our architecture reduces the error rate by 28.57% compared to prior art. We also propose a one-shot learning scenario of If-Then program synthesis and simulate it with our existing dataset. We demonstrate a variation on the training procedure for this scenario that outperforms the original procedure, significantly closing the gap to the model trained with all data.
Recent advances in machine learning have made significant contributions to drug discovery. Deep neural networks in particular have been demonstrated to provide significant boosts in predictive power when inferring the properties and activities of small-molecule compounds. However, the applicability of these techniques has been limited by the requirement for large amounts of training data. In this work, we demonstrate how one-shot learning can be used to significantly lower the amounts of data required to make meaningful predictions in drug discovery applications. We introduce a new architecture, the residual LSTM embedding, that, when combined with graph convolutional neural networks, significantly improves the ability to learn meaningful distance metrics over small-molecules. We open source all models introduced in this work as part of DeepChem, an open-source framework for deep-learning in drug discovery.
Generative adversarial networks (GANs) learn to synthesise new samples from a high-dimensional distribution by passing samples drawn from a latent space through a generative network. When the high-dimensional distribution describes images of a particular data set, the network should learn to generate visually similar image samples for latent variables that are close to each other in the latent space. For tasks such as image retrieval and image classification, it may be useful to exploit the arrangement of the latent space by projecting images into it, and using this as a representation for discriminative tasks. GANs often consist of multiple layers of non-linear computations, making them very difficult to invert. This paper introduces techniques for projecting image samples into the latent space using any pre-trained GAN, provided that the computational graph is available. We evaluate these techniques on both MNIST digits and Omniglot handwritten characters. In the case of MNIST digits, we show that projections into the latent space maintain information about the style and the identity of the digit. In the case of Omniglot characters, we show that even characters from alphabets that have not been seen during training may be projected well into the latent space; this suggests that this approach may have applications in one-shot learning.
Our world can be succinctly and compactly described as structured scenes of objects and relations. A typical room, for example, contains salient objects such as tables, chairs and books, and these objects typically relate to each other by their underlying causes and semantics. This gives rise to correlated features, such as position, function and shape. Humans exploit knowledge of objects and their relations for learning a wide spectrum of tasks, and more generally when learning the structure underlying observed data. In this work, we introduce relation networks (RNs) - a general purpose neural network architecture for object-relation reasoning. We show that RNs are capable of learning object relations from scene description data. Furthermore, we show that RNs can act as a bottleneck that induces the factorization of objects from entangled scene description inputs, and from distributed deep representations of scene images provided by a variational autoencoder. The model can also be used in conjunction with differentiable memory mechanisms for implicit relation discovery in one-shot learning tasks. Our results suggest that relation networks are a potentially powerful architecture for solving a variety of problems that require object relation reasoning.
Recent advances in one-shot learning have produced models that can learn from a handful of labeled examples, for passive classification and regression tasks. This paper combines reinforcement learning with one-shot learning, allowing the model to decide, during classification, which examples are worth labeling. We introduce a classification task in which a stream of images are presented and, on each time step, a decision must be made to either predict a label or pay to receive the correct label. We present a recurrent neural network based action-value function, and demonstrate its ability to learn how and when to request labels. Through the choice of reward function, the model can achieve a higher prediction accuracy than a similar model on a purely supervised task, or trade prediction accuracy for fewer label requests.
To have a robot actively supporting a human during a collaborative task, it is crucial that robots are able to identify the current action in order to predict the next one. Common approaches make use of high-level knowledge, such as object affordances, semantics or understanding of actions in terms of pre- and post-conditions. These approaches often require hand-coded a priori knowledge, time- and resource-intensive or supervised learning techniques. We propose to reframe this problem as an appearance-based place recognition problem. In our framework, we regard sequences of visual images of human actions as a map in analogy to the visual place recognition problem. Observing the task for the second time, our approach is able to recognize pre-observed actions in a one-shot learning approach and is thereby able to recognize the current observation in the task space. We propose two new methods for creating and aligning action observations within a task map. We compare and verify our approaches with real data of humans assembling several types of IKEA flat packs.
Despite recent advances, memory-augmented deep neural networks are still limited when it comes to life-long and one-shot learning, especially in remembering rare events. We present a large-scale life-long memory module for use in deep learning. The module exploits fast nearest-neighbor algorithms for efficiency and thus scales to large memory sizes. Except for the nearest-neighbor query, the module is fully differentiable and trained end-to-end with no extra supervision. It operates in a life-long manner, i.e., without the need to reset it during training. Our memory module can be easily added to any part of a supervised neural network. To show its versatility we add it to a number of networks, from simple convolutional ones tested on image classification to deep sequence-to-sequence and recurrent-convolutional models. In all cases, the enhanced network gains the ability to remember and do life-long one-shot learning. Our module remembers training examples shown many thousands of steps in the past and it can successfully generalize from them. We set new state-of-the-art for one-shot learning on the Omniglot dataset and demonstrate, for the first time, life-long one-shot learning in recurrent neural networks on a large-scale machine translation task.
Deep neural networks achieve unprecedented performance levels over many tasks and scale well with large quantities of data, but performance in the low-data regime and tasks like one shot learning still lags behind. While recent work suggests many hypotheses from better optimization to more complicated network structures, in this work we hypothesize that having a learnable and more expressive similarity objective is an essential missing component. Towards overcoming that, we propose a network design inspired by deep residual networks that allows the efficient computation of this more expressive pairwise similarity objective. Further, we argue that regularization is key in learning with small amounts of data, and propose an additional generator network based on the Generative Adversarial Networks where the discriminator is our residual pairwise network. This provides a strong regularizer by leveraging the generated data samples. The proposed model can generate plausible variations of exemplars over unseen classes and outperforms strong discriminative baselines for few shot classification tasks. Notably, our residual pairwise network design outperforms previous state-of-theart on the challenging mini-Imagenet dataset for one shot learning by getting over 55% accuracy for the 5-way classification task over unseen classes.
We present a novel cross-lingual transfer method for paradigm completion, the task of mapping a lemma to its inflected forms, using a neural encoder-decoder model, the state of the art for the monolingual task. We use labeled data from a high-resource language to increase performance on a low-resource language. In experiments on 21 language pairs from four different language families, we obtain up to 58% higher accuracy than without transfer and show that even zero-shot and one-shot learning are possible. We further find that the degree of language relatedness strongly influences the ability to transfer morphological knowledge.
We train a generator by maximum likelihood and we also train the same generator architecture by Wasserstein GAN. We then compare the generated samples, exact log-probability densities and approximate Wasserstein distances. We show that an independent critic trained to approximate Wasserstein distance between the validation set and the generator distribution helps detect overfitting. Finally, we use ideas from the one-shot learning literature to develop a novel fast learning critic.
Rotation invariance and translation invariance have great values in image recognition tasks. In this paper, we bring a new architecture in convolutional neural network (CNN) named cyclic convolutional layer to achieve rotation invariance in 2-D symbol recognition. We can also get the position and orientation of the 2-D symbol by the network to achieve detection purpose for multiple non-overlap target. Last but not least, this architecture can achieve one-shot learning in some cases using those invariance.
This paper aims at one-shot learning of deep neural nets, where a highly parallel setting is considered to address the algorithm calibration problem - selecting the best neural architecture and learning hyper-parameter values depending on the dataset at hand. The notoriously expensive calibration problem is optimally reduced by detecting and early stopping non-optimal runs. The theoretical contribution regards the optimality guarantees within the multiple hypothesis testing framework. Experimentations on the Cifar10, PTB and Wiki benchmarks demonstrate the relevance of the approach with a principled and consistent improvement on the state of the art with no extra hyper-parameter.