Faustino J. Gomez

Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

Many real-world sequence learning tasks require the prediction of sequences of labels from noisy, unsegmented input data. In speech recognition, for example, an acoustic signal is transcribed into words or sub-word units. Recurrent neural networks (RNNs) are powerful sequence learners that would seem well suited to such tasks. However, because they require pre-segmented training data, and post-processing to transform their outputs into label sequences, their applicability has so far been limited. This paper presents a novel method for training RNNs to label unsegmented sequences directly, thereby solving both problems. An experiment on the TIMIT speech corpus demonstrates its advantages over both a baseline HMM and a hybrid HMM-RNN.

Incremental evolution of complex general behavior

Several researchers have demonstrated how complex action sequences can be learned through neuroevolution (i.e., evolving neural networks with genetic algorithms). However, complex general behavior such as evading predators or avoiding obstacles, which is not tied to specific environments, turns out to be very difficult to evolve. Often the system discovers mechanical strategies, such as moving back and forth, that help the agent cope but are not very effective, do not appear believable, and do not generalize to new environments. The problem is that a general strategy is too difficult for the evolution system to discover directly. This article proposes an approach wherein such complex general behavior is learned incrementally, by starting with simpler behavior and gradually making the task more challenging and general. The task transitions are implemented through successive stages of Delta coding (i.e., evolving modifications), which allows even converged populations to adapt to the new task. The method is tested in the stochastic, dynamic task of prey capture and is compared with direct evolution. The incremental approach evolves more effective and more general behavior and should also scale up to harder tasks.

Training Recurrent Networks by Evolino

In recent years, gradient-based LSTM recurrent neural networks (RNNs) solved many previously RNN-unlearnable tasks. Sometimes, however, gradient information is of little use for training RNNs, due to numerous local minima. For such cases, we present a novel method: EVOlution of systems with LINear Outputs (Evolino). Evolino evolves weights to the nonlinear, hidden nodes of RNNs while computing optimal linear mappings from hidden state to output, using methods such as pseudo-inverse-based linear regression. If we instead use quadratic programming to maximize the margin, we obtain the first evolutionary recurrent support vector machines. We show that Evolino-based LSTM can solve tasks that Echo State nets (Jaeger, 2004a) cannot and achieves higher accuracy in certain continuous function generation tasks than conventional gradient descent RNNs, including gradient-based LSTM.

Efficient non-linear control through neuroevolution

Many complex control problems are not amenable to traditional controller design. Not only is it difficult to model real systems, but often it is unclear what kind of behavior is required. Reinforcement learning (RL) has made progress through direct interaction with the task environment, but it has been difficult to scale it up to large and partially observable state spaces. In recent years, neuroevolution, the artificial evolution of neural networks, has shown promise in tasks with these two properties. This paper introduces a novel neuroevolution method called CoSyNE that evolves networks at the level of weights. In the most extensive comparison of RL methods to date, it was tested in difficult versions of the pole-balancing problem that involve large state spaces and hidden state. CoSyNE was found to be significantly more efficient and powerful than the other methods on these tasks, forming a promising foundation for solving challenging real-world control tasks.

Sustaining diversity using behavioral information distance

Conventional similarity metrics used to sustain diversity in evolving populations are not well suited to sequential decision tasks. Genotypes and phenotypic structure are poor predictors of how solutions will actually behave in the environment. In this paper, we propose measuring similarity directly on the behavioral trajectories of evolving candidate policies using a universal similarity measure based on algorithmic information theory: normalized compression distance (NCD). NCD is compared to four other similarity measures in both genotype and phenotype space on the POMDP Tartarus problem, and shown to produce the most fit, general, and complex solutions.

Evolving large-scale neural networks for vision-based reinforcement learning

The idea of using evolutionary computation to train artificial neural networks, or neuroevolution (NE), for reinforcement learning (RL) tasks has now been around for over 20 years. However, as RL tasks become more challenging, the networks required become larger, as do their genomes. But, scaling NE to large nets (i.e. tens of thousands of weights) is infeasible using direct encodings that map genes one-to-one to network components. In this paper, we scale-up our compressed network encoding where network weight matrices are represented indirectly as a set of Fourier-type coefficients, to tasks that require very-large networks due to the high-dimensionality of their input space. The approach is demonstrated successfully on two reinforcement learning tasks in which the control networks receive visual input: (1) a vision-based version of the octopus control task requiring networks with over 3 thousand weights, and (2) a version of the TORCS driving game where networks with over 1 million weights are evolved to drive a car around a track using video images from the drivers perspective.

Co-evolving recurrent neurons learn deep memory POMDPs

Recurrent neural networks are theoretically capable of learning complex temporal sequences, but training them through gradient-descent is too slow and unstable for practical use in reinforcement learning environments. Neuroevolution, the evolution of artificial neural networks using genetic algorithms, can potentially solve real-world reinforcement learning tasks that require deep use of memory, i.e. memory spanning hundreds or thousands of inputs, by searching the space of recurrent neural networks directly. In this paper, we introduce a new neuroevolution algorithm called Hierarchical Enforced SubPopulations that simultaneously evolves networks at two levels of granularity: full networks and network components or neurons. We demonstrate the method in two POMDP tasks that involve temporal dependencies of up to thousands of time-steps, and show that it is faster and simpler than the current best conventional reinforcement learning system on these tasks.

Evolving deep unsupervised convolutional networks for vision-based reinforcement learning

Dealing with high-dimensional input spaces, like visual input, is a challenging task for reinforcement learning (RL). Neuroevolution (NE), used for continuous RL problems, has to either reduce the problem dimensionality by (1) compressing the representation of the neural network controllers or (2) employing a pre-processor (compressor) that transforms the high-dimensional raw inputs into low-dimensional features. In this paper, we are able to evolve extremely small recurrent neural network (RNN) controllers for a task that previously required networks with over a million weights. The high-dimensional visual input, which the controller would normally receive, is first transformed into a compact feature vector through a deep, max-pooling convolutional neural network (MPCNN). Both the MPCNN preprocessor and the RNN controller are evolved successfully to control a car in the TORCS racing simulator using only visual input. This is the first use of deep learning in the context evolutionary RL.

Modeling systems with internal state using evolino

Existing Recurrent Neural Networks (RNNs) are limited in their ability to model dynamical systems with nonlinearities and hidden internal states. Here we use our general framework for sequence learning, EVOlution of recurrent systems with LINear Outputs (Evolino), to discover good RNN hidden node weights through evolution, while using linear regression to compute an optimal linear mapping from hidden state to output. Using the Long Short-Term Memory RNN Architecture, Evolino outperforms previous state-of-the-art methods on several tasks: 1) context-sensitive languages, 2) multiple superimposed sine waves.

Planning to be surprised: optimal Bayesian exploration in dynamic environments

To maximize its success, an AGI typically needs to explore its initially unknown world. Is there an optimal way of doing so? Here we derive an affirmative answer for a broad class of environments.