Abstract-- Data-efficient learning of manipulation policies from visual observations is an outstanding challenge for realrobot learning. While deep reinforcement learning (RL) algorithms have shown success learning policies from visual observations, they still require an impractical number of real-world data samples to learn effective policies. However, recent advances in unsupervised representation learning and data augmentation significantly improved the sample efficiency of training RL policies on common simulated benchmarks. Building on these advances, we present a Framework for Efficient Robotic Manipulation (FERM) that utilizes data augmentation and unsupervised learning to achieve extremely sample-efficient training of robotic manipulation policies with sparse rewards. We show that, given only 10 demonstrations, a single robotic arm can learn sparse-reward manipulation policies from pixels, such as reaching, picking, moving, pulling a large object, flipping a switch, and opening a drawer in just 15-50 minutes of real-world training time.

Few-shot learning amounts to learning representations and acquiring knowledge such that novel tasks may be solved with both supervision and data being limited. Improved performance is possible by transductive inference, where the entire test set is available concurrently, and semi-supervised learning, where more unlabeled data is available. These problems are closely related because there is little or no adaptation of the representation in novel tasks. Focusing on these two settings, we introduce a new algorithm that leverages the manifold structure of the labeled and unlabeled data distribution to predict pseudo-labels, while balancing over classes and using the loss value distribution of a limited-capacity classifier to select the cleanest labels, iterately improving the quality of pseudo-labels. Our solution sets new state of the art on four benchmark datasets, namely \emph{mini}ImageNet, \emph{tiered}ImageNet, CUB and CIFAR-FS, while being robust over feature space pre-processing and the quantity of available data.

Monte Carlo tree search (MCTS) has achieved state-of-the-art results in many domains such as Go and Atari games when combining with deep neural networks (DNNs). When more simulations are executed, MCTS can achieve higher performance but also requires enormous amounts of CPU and GPU resources. However, not all states require a long searching time to identify the best action that the agent can find. For example, in 19x19 Go and NoGo, we found that for more than half of the states, the best action predicted by DNN remains unchanged even after searching 2 minutes. This implies that a significant amount of resources can be saved if we are able to stop the searching earlier when we are confident with the current searching result. In this paper, we propose to achieve this goal by predicting the uncertainty of the current searching status and use the result to decide whether we should stop searching. With our algorithm, called Dynamic Simulation MCTS (DS-MCTS), we can speed up a NoGo agent trained by AlphaZero 2.5 times faster while maintaining a similar winning rate. Also, under the same average simulation count, our method can achieve a 61% winning rate against the original program.

Many neural network models have been successful at classification problems, but their operation is still treated as a black box. Here, we developed a theory for one-layer perceptrons that can predict performance on classification tasks. This theory is a generalization of an existing theory for predicting the performance of Echo State Networks and connectionist models for symbolic reasoning known as Vector Symbolic Architectures. In this paper, we first show that the proposed perceptron theory can predict the performance of Echo State Networks, which could not be described by the previous theory. Second, we apply our perceptron theory to the last layers of shallow randomly connected and deep multi-layer networks. The full theory is based on Gaussian statistics, but it is analytically intractable. We explore numerical methods to predict network performance for problems with a small number of classes. For problems with a large number of classes, we investigate stochastic sampling methods and a tractable approximation to the full theory. The quality of predictions is assessed in three experimental settings, using reservoir computing networks on a memorization task, shallow randomly connected networks on a collection of classification datasets, and deep convolutional networks with the ImageNet dataset. This study offers a simple, bipartite approach to understand deep neural networks: the input is encoded by the last-but-one layers into a high-dimensional representation. This representation is mapped through the weights of the last layer into the postsynaptic sums of the output neurons. Specifically, the proposed perceptron theory uses the mean vector and covariance matrix of the postsynaptic sums to compute classification accuracies for the different classes. The first two moments of the distribution of the postsynaptic sums can predict the overall network performance quite accurately.

Traditional Optical Character Recognition (OCR) systems that generate text of highly inflectional Indic languages like Hindi tend to suffer from poor accuracy due to a wide alphabet set, compound characters and difficulty in segmenting characters in a word. Automatic spelling error detection and context-sensitive error correction can be used to improve accuracy by post-processing the text generated by these OCR systems. A majority of previously developed language models for error correction of Hindi spelling have been context-free. In this paper, we present Vartani Spellcheck - a context-sensitive approach for spelling correction of Hindi text using a state-of-the-art transformer - BERT in conjunction with the Levenshtein distance algorithm, popularly known as Edit Distance. We use a lookup dictionary and context-based named entity recognition (NER) for detection of possible spelling errors in the text. Our proposed technique has been tested on a large corpus of text generated by the widely used Tesseract OCR on the Hindi epic Ramayana. With an accuracy of 81%, the results show a significant improvement over some of the previously established context-sensitive error correction mechanisms for Hindi. We also explain how Vartani Spellcheck may be used for on-the-fly autocorrect suggestion during continuous typing in a text editor environment.

In-home health monitoring has attracted great attention for the ageing population worldwide. With the abundant user health data accessed by Internet of Things (IoT) devices and recent development in machine learning, smart healthcare has seen many successful stories. However, existing approaches for in-home health monitoring do not pay sufficient attention to user data privacy and thus are far from being ready for large-scale practical deployment. In this paper, we propose FedHome, a novel cloud-edge based federated learning framework for in-home health monitoring, which learns a shared global model in the cloud from multiple homes at the network edges and achieves data privacy protection by keeping user data locally. To cope with the imbalanced and non-IID distribution inherent in user's monitoring data, we design a generative convolutional autoencoder (GCAE), which aims to achieve accurate and personalized health monitoring by refining the model with a generated class-balanced dataset from user's personal data. Besides, GCAE is lightweight to transfer between the cloud and edges, which is useful to reduce the communication cost of federated learning in FedHome. Extensive experiments based on realistic human activity recognition data traces corroborate that FedHome significantly outperforms existing widely-adopted methods.

Segmentation of findings in the gastrointestinal tract is a challenging but also an important task which is an important building stone for sufficient automatic decision support systems. In this work, we present our solution for the Medico 2020 task, which focused on the problem of colon polyp segmentation. We present our simple but efficient idea of using an augmentation method that uses grids in a pyramid-like manner (large to small) for segmentation. Our results show that the proposed methods work as indented and can also lead to comparable results when competing with other methods.

Ultra-lightweight model design is an important topic for the deployment of existing speech enhancement and source separation techniques on low-resource platforms. Various lightweight model design paradigms have been proposed in recent years; however, most models still suffer from finding a balance between model size, model complexity, and model performance. In this paper, we propose the group communication with context codec (GC3) design to decrease both model size and complexity without sacrificing the model performance. Group communication splits a high-dimensional feature into groups of low-dimensional features and applies a module to capture the inter-group dependency. A model can then be applied to the groups in parallel with a significantly smaller width. A context codec is applied to decrease the length of a sequential feature, where a context encoder compresses the temporal context of local features into a single feature representing the global characteristics of the context, and a context decoder decompresses the transformed global features back to the context features. Experimental results show that GC3 can achieve on par or better performance than a wide range of baseline architectures with as small as 2.5% model size.

Human intelligence is characterized by a remarkable ability to infer abstract rules from experience and apply these rules to novel domains. As such, designing neural network algorithms with this capacity is an important step toward the development of deep learning systems with more human-like intelligence. However, doing so is a major outstanding challenge, one that some argue will require neural networks to use explicit symbol-processing mechanisms. In this work, we focus on neural networks' capacity for arbitrary role-filler binding, the ability to associate abstract "roles" to context-specific "fillers," which many have argued is an important mechanism underlying the ability to learn and apply rules abstractly. Using a simplified version of Raven's Progressive Matrices, a hallmark test of human intelligence, we introduce a sequential formulation of a visual problem-solving task that requires this form of binding. Further, we introduce the Emergent Symbol Binding Network (ESBN), a recurrent neural network model that learns to use an external memory as a binding mechanism. This mechanism enables symbol-like variable representations to emerge through the ESBN's training process without the need for explicit symbol-processing machinery. We empirically demonstrate that the ESBN successfully learns the underlying abstract rule structure of our task and perfectly generalizes this rule structure to novel fillers.

Current voice assistants typically use the best hypothesis yielded by their Automatic Speech Recognition (ASR) module as input to their Natural Language Understanding (NLU) module, thereby losing helpful information that might be stored in lower-ranked ASR hypotheses. We explore the change in performance of NLU associated tasks when utilizing five-best ASR hypotheses when compared to status quo for two language datasets, German and Portuguese. To harvest information from the ASR five-best, we leverage extractive summarization and joint extractive-abstractive summarization models for Domain Classification (DC) experiments while using a sequence-to-sequence model with a pointer generator network for Intent Classification (IC) and Named Entity Recognition (NER) multi-task experiments. For the DC full test set, we observe significant improvements of up to 7.2% and 15.5% in micro-averaged F1 scores, for German and Portuguese, respectively. In cases where the best ASR hypothesis was not an exact match to the transcribed utterance (mismatched test set), we see improvements of up to 6.7% and 8.8% micro-averaged F1 scores, for German and Portuguese, respectively. For IC and NER multi-task experiments, when evaluating on the mismatched test set, we see improvements across all domains in German and in 17 out of 19 domains in Portuguese (improvements based on change in SeMER scores). Our results suggest that the use of multiple ASR hypotheses, as opposed to one, can lead to significant performance improvements in the DC task for these non-English datasets. In addition, it could lead to significant improvement in the performance of IC and NER tasks in cases where the ASR model makes mistakes.