Animal vocalization denoising is a task similar to human speech enhancement, which is relatively well-studied. In contrast to the latter, it comprises a higher diversity of sound production mechanisms and recording environments, and this higher diversity is a challenge for existing models. Adding to the challenge and in contrast to speech, we lack large and diverse datasets comprising clean vocalizations. As a solution we use as training data pseudo-clean targets, i.e. pre-denoised vocalizations, and segments of background noise without a vocalization. We propose a train set derived from bioacoustics datasets and repositories representing diverse species, acoustic environments, geographic regions. Additionally, we introduce a non-overlapping benchmark set comprising clean vocalizations from different taxa and noise samples. We show that that denoising models (demucs, CleanUNet) trained on pseudo-clean targets obtained with speech enhancement models achieve competitive results on the benchmarking set. We publish data, code, libraries, and demos at https://mariusmiron.com/research/biodenoising.

Online and offline $\mathtt{RLHF}$ methods, such as $\mathtt{PPO}$ and $\mathtt{DPO}$, have been highly successful in aligning AI with human preferences. Despite their success, however, these methods suffer from fundamental limitations: $\mathbf{(a)}$ Models trained with $\mathtt{RLHF}$ can learn from mistakes or negative examples through RL mechanism or contrastive loss during training. However, at inference time, they lack an innate self-improvement mechanism for error corrections. $\mathbf{(b)}$ The optimal solution of existing methods is highly task-dependent, making it difficult for them to generalize to new tasks. To address these challenges, we propose Self-Improving Robust Preference Optimization ($\mathtt{SRPO}$), a practical and mathematically principled offline $\mathtt{RLHF}$ framework. The key idea behind $\mathtt{SRPO}$ is to cast the problem of learning from human preferences as a self-improvement process, mathematically formulated as a min-max objective that jointly optimizes a self-improvement policy and a generative policy in an adversarial fashion. Crucially, the solution for this optimization problem is independent of the training task, which makes it robust to its changes. We then show that this objective can be reformulated as a non-adversarial offline loss, which can be efficiently optimized using standard supervised learning techniques at scale. To demonstrate $\mathtt{SRPO}$’s effectiveness, we evaluate it using AI Win-Rate (WR) against human (GOLD) completions. When tested on the XSum dataset, $\mathtt{SRPO}$ outperforms $\mathtt{DPO}$ by a margin of $\mathbf{15}$% after $5$ self-revisions, achieving an impressive $\mathbf{90}$% WR. Moreover, on the challenging Arena-Hard prompts, $\mathtt{SRPO}$ outperforms both $\mathtt{DPO}$ and $\mathtt{IPO}$ (by $\mathbf{4}$% without revision and $\mathbf{6}$% after a single revision), reaching a $\mathbf{56}$% WR against against $\mathtt{Llama-3.1-8B-Instruct}$.

Large language models (LLMs) prompted with text and audio have achieved state-of-the-art performance across various auditory tasks, including speech, music, and general audio, showing emergent abilities on unseen tasks. However, their potential has yet to be fully demonstrated in bioacoustics tasks, such as detecting animal vocalizations in large recordings, classifying rare and endangered species, and labeling context and behavior—tasks that are crucial for conservation, biodiversity monitoring, and animal behavior studies. In this work, we present NatureLM-audio, the first audio-language foundation model specifically designed for bioacoustics. Our training dataset consists of carefully curated text-audio pairs spanning bioacoustics, speech, and music, designed to address the field's limited availability of annotated data. We demonstrate successful transfer of learned representations from music and speech to bioacoustics, and our model shows promising generalization to unseen taxa and tasks. We evaluate NatureLM-audio on a novel benchmark (BEANS-Zero) and it sets a new state of the art on several bioacoustics tasks, including zero-shot classification of unseen species. To advance bioacoustics research, we release our model weights, benchmark data, and open-source the code for training and benchmark data generation and model training.

The fine-tuning of pre-trained large language models (LLMs) using reinforcement learning (RL) is generally formulated as direct policy optimization. This approach was naturally favored as it efficiently improves a pretrained LLM with simple gradient updates. Another RL paradigm, Q-learning methods, has received far less attention in the LLM community while demonstrating major success in various non-LLM RL tasks. In particular, Q-learning effectiveness stems from its sample efficiency and ability to learn offline, which is particularly valuable given the high computational cost of sampling with LLM. However, naively applying a Q-learning–style update to the model’s logits is ineffective due to the specificity of LLMs. Our contribution is to derive theoretically grounded loss functions from Bellman equations to adapt Q-learning methods to LLMs. To do so, we interpret LLM logits as Q-values and carefully adapt insights from the RL literature to account for LLM-specific characteristics. It thereby ensures that the logits become reliable Q-value estimates. We then use this loss to build a practical algorithm, ShiQ for Shifted-Q, that supports off-policy, token-wise learning while remaining simple to implement. Finally, ShiQ is evaluated on both synthetic data and real-world benchmarks, e.g., UltraFeedback, BFCL-V3, demonstrating its effectiveness in both single-turn and multi-turn LLM settings.

Recent techniques based on Mean Field Games (MFGs) allow the scalable analysis of multi-player games with many similar, rational agents. However, standard MFGs remain limited to homogeneous players that weakly influence each other, and cannot model major players that strongly influence other players, severely limiting the class of problems that can be handled. We propose a novel discrete time version of major-minor MFGs (M3FGs), along with a learning algorithm based on fictitious play and partitioning the probability simplex. Importantly, M3FGs generalize MFGs with common noise and can handle not only random exogeneous environment states but also major players. A key challenge is that the mean field is stochastic and not deterministic as in standard MFGs. Our theoretical investigation verifies both the M3FG model and its algorithmic solution, showing firstly the well-posedness of the M3FG model starting from a finite game of interest, and secondly convergence and approximation guarantees of the fictitious play algorithm. Then, we empirically verify the obtained theoretical results, ablating some of the theoretical assumptions made, and show successful equilibrium learning in three example problems. Overall, we establish a learning framework for a novel and broad class of tractable games.

Arash Ahmadian, Chris Cremer, Matthias Gallé, Marzieh Fadaee, Julia Kreutzer, Olivier Pietquin, Ahmet Üstün, Sara Hooker. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Yannis Flet-Berliac, Nathan Grinsztajn, Florian Strub, Eugene Choi, Bill Wu, Chris Cremer, Arash Ahmadian, Yash Chandak, Mohammad Gheshlaghi Azar, Olivier Pietquin, Matthieu Geist. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

We propose MusicRL, the first music generation system finetuned from human feedback. Appreciation of text-to-music models is particularly subjective since the concept of musicality as well as the specific intention behind a caption are user-dependent (e.g. a caption such as “upbeat workout music” can map to a retro guitar solo or a technopop beat). Not only this makes supervised training of such models challenging, but it also calls for integrating continuous human feedback in their post-deployment finetuning. MusicRL is a pretrained autoregressive MusicLM model of discrete audio tokens finetuned with reinforcement learning to maximize sequence-level rewards. We design reward functions related specifically to text-adherence and audio quality with the help from selected raters, and use those to finetune MusicLM into MusicRL-R. We deploy MusicLM to users and collect a substantial dataset comprising 300,000 pairwise preferences. Using Reinforcement Learning from Human Feedback (RLHF), we train MusicRL-U, the first text-to-music model that incorporates human feedback at scale. Human evaluations show that both MusicRL-R and MusicRL-U are preferred to the baseline. Ultimately, MusicRL-RU combines the two approaches and results in the best model according to human raters. Ablation studies shed light on the musical attributes influencing human preferences, indicating that text adherence and quality only account for a part of it. This underscores the prevalence of subjectivity in musical appreciation and calls for further involvement of human listeners in the finetuning of music generation models. Samples can be found at google-research.github.io/seanet/musiclm/rlhf/.

Paul Roit, Johan Ferret, Lior Shani, Roee Aharoni, Geoffrey Cideron, Robert Dadashi, Matthieu Geist, Sertan Girgin, Leonard Hussenot, Orgad Keller, Nikola Momchev, Sabela Ramos Garea, Piotr Stanczyk, Nino Vieillard, Olivier Bachem, Gal Elidan, Avinatan Hassidim, Olivier Pietquin, Idan Szpektor. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Mirror descent value iteration (MDVI), an abstraction of Kullback-Leibler (KL) and entropy-regularized reinforcement learning (RL), has served as the basis for recent high-performing practical RL algorithms. However, despite the use of function approximation in practice, the theoretical understanding of MDVI has been limited to tabular Markov decision processes (MDPs). We study MDVI with linear function approximation through its sample complexity required to identify an $\varepsilon$-optimal policy with probability $1-\delta$ under the settings of an infinite-horizon linear MDP, generative model, and G-optimal design. We demonstrate that least-squares regression weighted by the variance of an estimated optimal value function of the next state is crucial to achieving minimax optimality. Based on this observation, we present Variance-Weighted Least-Squares MDVI (VWLS-MDVI), the first theoretical algorithm that achieves nearly minimax optimal sample complexity for infinite-horizon linear MDPs. Furthermore, we propose a practical VWLS algorithm for value-based deep RL, Deep Variance Weighting (DVW). Our experiments demonstrate that DVW improves the performance of popular value-based deep RL algorithms on a set of MinAtar benchmarks.

We explore the problem of imitation learning (IL) in the context of mean-field games (MFGs), where the goal is to imitate the behavior of a population of agents following a Nash equilibrium policy according to some unknown payoff function. IL in MFGs presents new challenges compared to single-agent IL, particularly when both the reward function and the transition kernel depend on the population distribution. In this paper, departing from the existing literature on IL for MFGs, we introduce a new solution concept called the Nash imitation gap. Then we show that when only the reward depends on the population distribution, IL in MFGs can be reduced to single-agent IL with similar guarantees. However, when the dynamics is population-dependent, we provide a novel upper-bound that suggests IL is harder in this setting. To address this issue, we propose a new adversarial formulation where the reinforcement learning problem is replaced by a mean-field control (MFC) problem, suggesting progress in IL within MFGs may have to build upon MFC.

Abstract We introduce SPEAR-TTS, a multi-speaker text-to-speech (TTS) system that can be trained with minimal supervision. By combining two types of discrete speech representations, we cast TTS as a composition of two sequence-to-sequence tasks: from text to high-level semantic tokens (akin to “reading”) and from semantic tokens to low-level acoustic tokens (“speaking”). Decoupling these two tasks enables training of the “speaking” module using abundant audio-only data, and unlocks the highly efficient combination of pretraining and backtranslation to reduce the need for parallel data when training the “reading” component. To control the speaker identity, we adopt example prompting, which allows SPEAR-TTS to generalize to unseen speakers using only a short sample of 3 seconds, without any explicit speaker representation or speaker labels. Our experiments demonstrate that SPEAR-TTS achieves a character error rate that is competitive with state-of-the-art methods using only 15 minutes of parallel data, while matching ground-truth speech in naturalness and acoustic quality.

We introduce AudioLM, a framework for high-quality audio generation with long-term consistency. AudioLM maps the input audio to a sequence of discrete tokens and casts audio generation as a language modeling task in this representation space. We show how existing audio tokenizers provide different trade-offs between reconstruction quality and long-term structure, and we propose a hybrid tokenization scheme to achieve both objectives. Namely, we leverage the discretized activations of a masked language model pre-trained on audio to capture long-term structure and the discrete codes produced by a neural audio codec to achieve high-quality synthesis. By training on large corpora of raw audio waveforms, AudioLM learns to generate natural and coherent continuations given short prompts. When trained on speech, and without any transcript or annotation, AudioLM generates syntactically and semantically plausible speech continuations while also maintaining speaker identity and prosody for unseen speakers. Furthermore, we demonstrate how our approach extends beyond speech by generating coherent piano music continuations, despite being trained without any symbolic representation of music.

Mean Field Games (MFGs) can potentially scale multi-agent systems to extremely large populations of agents. Yet, most of the literature assumes a single initial distribution for the agents, which limits the practical applications of MFGs. Machine Learning has the potential to solve a wider diversity of MFG problems thanks to generalizations capacities. We study how to leverage these generalization properties to learn policies enabling a typical agent to behave optimally against any population distribution. In reference to the Master equation in MFGs, we coin the term “Master policies” to describe them and we prove that a single Master policy provides a Nash equilibrium, whatever the initial distribution. We propose a method to learn such Master policies. Our approach relies on three ingredients: adding the current population distribution as part of the observation, approximating Master policies with neural networks, and training via Reinforcement Learning and Fictitious Play. We illustrate on numerical examples not only the efficiency of the learned Master policy but also its generalization capabilities beyond the distributions used for training.

Offline Reinforcement Learning (RL) aims at learning an optimal control from a fixed dataset, without interactions with the system. An agent in this setting should avoid selecting actions whose consequences cannot be predicted from the data. This is the converse of exploration in RL, which favors such actions. We thus take inspiration from the literature on bonus-based exploration to design a new offline RL agent. The core idea is to subtract a prediction-based exploration bonus from the reward, instead of adding it for exploration. This allows the policy to stay close to the support of the dataset and practically extends some previous pessimism-based offline RL methods to a deep learning setting with arbitrary bonuses. We also connect this approach to a more common regularization of the learned policy towards the data. Instantiated with a bonus based on the prediction error of a variational autoencoder, we show that our simple agent is competitive with the state of the art on a set of continuous control locomotion and manipulation tasks.

The $Q$-function is a central quantity in many Reinforcement Learning (RL) algorithms for which RL agents behave following a (soft)-greedy policy w.r.t. to $Q$. It is a powerful tool that allows action selection without a model of the environment and even without explicitly modeling the policy. Yet, this scheme can only be used in discrete action tasks, with small numbers of actions, as the softmax over actions cannot be computed exactly otherwise. More specifically, the usage of function approximation to deal with continuous action spaces in modern actor-critic architectures intrinsically prevents the exact computation of a softmax. We propose to alleviate this issue by parametrizing the $Q$-function implicitly, as the sum of a log-policy and a value function. We use the resulting parametrization to derive a practical off-policy deep RL algorithm, suitable for large action spaces, and that enforces the softmax relation between the policy and the $Q$-value. We provide a theoretical analysis of our algorithm: from an Approximate Dynamic Programming perspective, we show its equivalence to a regularized version of value iteration, accounting for both entropy and Kullback-Leibler regularization, and that enjoys beneficial error propagation results. We then evaluate our algorithm on classic control tasks, where its results compete with state-of-the-art methods.

In this paper, we propose a novel Reinforcement Learning (RL) framework for problems with continuous action spaces: Action Quantization from Demonstrations (AQuaDem). The proposed approach consists in learning a discretization of continuous action spaces from human demonstrations. This discretization returns a set of plausible actions (in light of the demonstrations) for each input state, thus capturing the priors of the demonstrator and their multimodal behavior. By discretizing the action space, any discrete action deep RL technique can be readily applied to the continuous control problem. Experiments show that the proposed approach outperforms state-of-the-art methods such as SAC in the RL setup, and GAIL in the Imitation Learning setup. We provide a website with interactive videos: https://google-research.github.io/aquadem/ and make the code available: https://github.com/google-research/google-research/tree/master/aquadem.

Mean Field Games (MFGs) have been introduced to efficiently approximate games with very large populations of strategic agents. Recently, the question of learning equilibria in MFGs has gained momentum, particularly using model-free reinforcement learning (RL) methods. One limiting factor to further scale up using RL is that existing algorithms to solve MFGs require the mixing of approximated quantities such as strategies or $q$-values. This is far from being trivial in the case of non-linear function approximation that enjoy good generalization properties, e.g. neural networks. We propose two methods to address this shortcoming. The first one learns a mixed strategy from distillation of historical data into a neural network and is applied to the Fictitious Play algorithm. The second one is an online mixing method based on regularization that does not require memorizing historical data or previous estimates. It is used to extend Online Mirror Descent. We demonstrate numerically that these methods efficiently enable the use of Deep RL algorithms to solve various MFGs. In addition, we show that these methods outperform SotA baselines from the literature.

Alice Martin, Guillaume Quispe, Charles Ollion, Sylvain Le Corff, Florian Strub, Olivier Pietquin. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Lewis signaling games are a class of simple communication games for simulating the emergence of language. In these games, two agents must agree on a communication protocol in order to solve a cooperative task. Previous work has shown that agents trained to play this game with reinforcement learning tend to develop languages that display undesirable properties from a linguistic point of view (lack of generalization, lack of compositionality, etc). In this paper, we aim to provide better understanding of this phenomenon by analytically studying the learning problem in Lewis games. As a core contribution, we demonstrate that the standard objective in Lewis games can be decomposed in two components: a co-adaptation loss and an information loss. This decomposition enables us to surface two potential sources of overfitting, which we show may undermine the emergence of a structured communication protocol. In particular, when we control for overfitting on the co-adaptation loss, we recover desired properties in the emergent languages: they are more compositional and generalize better.

We propose CHARM, a method for training a single neural network across inconsistent input channels. Our work is motivated by Electroencephalography (EEG), where data collection protocols from different headsets result in varying channel ordering and number, which limits the feasibility of transferring trained systems across datasets. Our approach builds upon attention mechanisms to estimate a latent reordering matrix from each input signal and map input channels to a canonical order. CHARM is differentiable and can be composed further with architectures expecting a consistent channel ordering to build end-to-end trainable classifiers. We perform experiments on four EEG classification datasets and demonstrate the efficacy of CHARM via simulated shuffling and masking of input channels. Moreover, our method improves the transfer of pre-trained representations between datasets collected with different protocols.

Offline Reinforcement Learning methods seek to learn a policy from logged transitions of an environment, without any interaction. In the presence of function approximation, and under the assumption of limited coverage of the state-action space of the environment, it is necessary to enforce the policy to visit state-action pairs close to the support of logged transitions. In this work, we propose an iterative procedure to learn a pseudometric (closely related to bisimulation metrics) from logged transitions, and use it to define this notion of closeness. We show its convergence and extend it to the function approximation setting. We then use this pseudometric to define a new lookup based bonus in an actor-critic algorithm: PLOFF. This bonus encourages the actor to stay close, in terms of the defined pseudometric, to the support of logged transitions. Finally, we evaluate the method on hand manipulation and locomotion tasks.

We address the issue of tuning hyperparameters (HPs) for imitation learning algorithms in the context of continuous-control, when the underlying reward function of the demonstrating expert cannot be observed at any time. The vast literature in imitation learning mostly considers this reward function to be available for HP selection, but this is not a realistic setting. Indeed, would this reward function be available, it could then directly be used for policy training and imitation would not be necessary. To tackle this mostly ignored problem, we propose a number of possible proxies to the external reward. We evaluate them in an extensive empirical study (more than 10’000 agents across 9 environments) and make practical recommendations for selecting HPs. Our results show that while imitation learning algorithms are sensitive to HP choices, it is often possible to select good enough HPs through a proxy to the reward function.

We present a method enabling a large number of agents to learn how to flock. This problem has drawn a lot of interest but requires many structural assumptions and is tractable only in small dimensions. We phrase this problem as a Mean Field Game (MFG), where each individual chooses its own acceleration depending on the population behavior. Combining Deep Reinforcement Learning (RL) and Normalizing Flows (NF), we obtain a tractable solution requiring only very weak assumptions. Our algorithm finds a Nash Equilibrium and the agents adapt their velocity to match the neighboring flock’s average one. We use Fictitious Play and alternate: (1) computing an approximate best response with Deep RL, and (2) estimating the next population distribution with NF. We show numerically that our algorithm can learn multi-group or high-dimensional flocking with obstacles.

Sparse rewards are double-edged training signals in reinforcement learning: easy to design but hard to optimize. Intrinsic motivation guidances have thus been developed toward alleviating the resulting exploration problem. They usually incentivize agents to look for new states through novelty signals. Yet, such methods encourage exhaustive exploration of the state space rather than focusing on the environment's salient interaction opportunities. We propose a new exploration method, called Don't Do What Doesn't Matter (DoWhaM), shifting the emphasis from state novelty to state with relevant actions. While most actions consistently change the state when used, e.g. moving the agent, some actions are only effective in specific states, e.g., opening a door, grabbing an object. DoWhaM detects and rewards actions that seldom affect the environment. We evaluate DoWhaM on the procedurally-generated environment MiniGrid against state-of-the-art methods. Experiments consistently show that DoWhaM greatly reduces sample complexity, installing the new state-of-the-art in MiniGrid.

We propose to learn to distinguish reversible from irreversible actions for better informed decision-making in Reinforcement Learning (RL). From theoretical considerations, we show that approximate reversibility can be learned through a simple surrogate task: ranking randomly sampled trajectory events in chronological order. Intuitively, pairs of events that are always observed in the same order are likely to be separated by an irreversible sequence of actions. Conveniently, learning the temporal order of events can be done in a fully self-supervised way, which we use to estimate the reversibility of actions from experience, without any priors.We propose two different strategies that incorporate reversibility in RL agents, one strategy for exploration (RAE) and one strategy for control (RAC). We demonstrate the potential of reversibility-aware agents in several environments, including the challenging Sokoban game. In synthetic tasks, we show that we can learn control policies that never fail and reduce to zero the side-effects of interactions, even without access to the reward function.

Adversarial imitation learning has become a popular framework for imitation in continuous control. Over the years, several variations of its components were proposed to enhance the performance of the learned policies as well as the sample complexity of the algorithm. In practice, these choices are rarely tested all together in rigorous empirical studies.It is therefore difficult to discuss and understand what choices, among the high-level algorithmic options as well as low-level implementation details, matter. To tackle this issue, we implement more than 50 of these choices in a generic adversarial imitation learning frameworkand investigate their impacts in a large-scale study (>500k trained agents) with both synthetic and human-generated demonstrations. We analyze the key results and highlight the most surprising findings.

Learning by experience in Multi-Agent Systems (MAS) is a difficult and exciting task, due to the lack of stationarity of the environment, whose dynamics evolves as the population learns. In order to design scalable algorithms for systems with a large population of interacting agents (e.g., swarms), this paper focuses on Mean Field MAS, where the number of agents is asymptotically infinite. Recently, a very active burgeoning field studies the effects of diverse reinforcement learning algorithms for agents with no prior information on a stationary Mean Field Game (MFG) and learn their policy through repeated experience. We adopt a high perspective on this problem and analyze in full generality the convergence of a fictitious iterative scheme using any single agent learning algorithm at each step. We quantify the quality of the computed approximate Nash equilibrium, in terms of the accumulated errors arising at each learning iteration step. Notably, we show for the first time convergence of model free learning algorithms towards non-stationary MFG equilibria, relying only on classical assumptions on the MFG dynamics. We illustrate our theoretical results with a numerical experiment in a continuous action-space environment, where the approximate best response of the iterative fictitious play scheme is computed with a deep RL algorithm.

Conservative Policy Iteration (CPI) is a founding algorithm of Approximate Dynamic Programming (ADP). Its core principle is to stabilize greediness through stochastic mixtures of consecutive policies. It comes with strong theoretical guarantees, and inspired approaches in deep Reinforcement Learning (RL). However, CPI itself has rarely been implemented, never with neural networks, and only experimented on toy problems. In this paper, we show how CPI can be practically combined with deep RL with discrete actions, in an off-policy manner. We also introduce adaptive mixture rates inspired by the theory. We experiment thoroughly the resulting algorithm on the simple Cartpole problem, and validate the proposed method on a representative subset of Atari games. Overall, this work suggests that revisiting classic ADP may lead to improved and more stable deep RL algorithms.

This paper extends the notion of learning algorithms and learning equilibriums from repeated games theory to stochastic games. We introduce Foolproof Cooperative Learning (FCL), an algorithm that converges to an equilibrium strategy that allows cooperative strategies in self-play setting while being not exploitable by selfish learners. By construction, FCL is a learning equilibrium for repeated symmetric games. We illustrate the behavior of FCL on symmetric matrix and grid games, and its robustness to selfish learners.

We adapt the optimization’s concept of momentum to reinforcement learning. Seeing the state-action value functions as an anlog to the gradients in optimization, we interpret momentum as an average of consecutive $q$-functions. We derive Momentum Value Iteration (MoVI), a variation of Value iteration that incorporates this momentum idea. Our analysis shows that this allows MoVI to average errors over successive iterations. We show that the proposed approach can be readily extended to deep learning. Specifically,we propose a simple improvement on DQN based on MoVI, and experiment it on Atari games.

Language drift has been one of the major obstacles to train language models through interaction.When word-based conversational agents are trained towards completing a task, they tend to invent their language rather than leveraging natural language.In recent literature, two general methods partially counter this phenomenon: Supervised Selfplay (S2P) and Seeded Iterated Learning (SIL).While S2P jointly trains interactive and supervised losses to counter the drift, SIL changes the training dynamics to prevent language drift from occurring.In this paper, we first highlight their respective weaknesses, i.e., latestage training collapses and higher negative likelihood when evaluated on human corpus.Given these observations, we introduce Supervised Seeded Iterated Learning (SSIL) to combine both methods to minimize their respective weaknesses.We then show the effectiveness of SSIL in the language-drift translation game.

Pretraining on human corpus and then finetuning in a simulator has become a standard pipeline for training a goal-oriented dialogue agent. Nevertheless, as soon as the agents are finetuned to maximize task completion, they suffer from the so-called language drift phenomenon: they slowly lose syntactic and semantic properties of language as they only focus on solving the task. In this paper, we propose a generic approach to counter language drift called Seeded iterated learning (SIL). We periodically refine a pretrained student agent by imitating data sampled from a newly generated teacher agent. At each time step, the teacher is created by copying the student agent, before being finetuned to maximize task completion. SIL does not require external syntactic constraint nor semantic knowledge, making it a valuable task-agnostic finetuning protocol. We evaluate SIL in a toy-setting Lewis Game, and then scale it up to the translation game with natural language. In both settings, SIL helps counter language drift as well as it improves the task completion compared to baselines.

The ability to transfer knowledge to novel environments and tasks is a sensible desiderata for general learning agents. Despite the apparent promises, transfer in RL is still an open and little exploited research area. In this paper, we take a brand-new perspective about transfer: we suggest that the ability to assign credit unveils structural invariants in the tasks that can be transferred to make RL more sample-efficient. Our main contribution is SECRET, a novel approach to transfer learning for RL that uses a backward-view credit assignment mechanism based on a self-attentive architecture. Two aspects are key to its generality: it learns to assign credit as a separate offline supervised process and exclusively modifies the reward function. Consequently, it can be supplemented by transfer methods that do not modify the reward function and it can be plugged on top of any RL algorithm.

The use of Reinforcement Learning (RL) is still restricted to simulation or to enhance human-operated systems through recommendations. Real-world environments (e.g. industrial robots or power grids) are generally designed with safety constraints in mind implemented in the shape of valid actions masks or contingency controllers. For example, the range of motion and the angles of the motors of a robot can be limited to physical boundaries. Violating constraints thus results in rejected actions or entering in a safe mode driven by an external controller, making RL agents incapable of learning from their mistakes. In this paper, we propose a simple modification of a state-of-the-art deep RL algorithm (DQN), enabling learning from forbidden actions. To do so, the standard Q-learning update is enhanced with an extra safety loss inspired by structured classification. We empirically show that it reduces the number of hit constraints during the learning phase and accelerates convergence to near-optimal policies compared to using standard DQN. Experiments are done on a Visual Grid World Environment and the TextWorld domain.

Speaker recognition is a well known and studied task in the speech processing domain. It has many applications, either for security or speaker adaptation of personal devices. In this paper, we present a new paradigm for automatic speaker recognition that we call Interactive Speaker Recognition (ISR). In this paradigm, the recognition system aims to incrementally build a representation of the speakers by requesting personalized utterances to be spoken in contrast to the standard text-dependent or text-independent schemes. To do so, we cast the speaker recognition task into a sequential decision-making problem that we solve with Reinforcement Learning. Using a standard dataset, we show that our method achieves excellent performance while using little speech signal amounts. This method could also be applied as an utterance selection mechanism for building speech synthesis systems.

In this paper, we deepen the analysis of continuous time Fictitious Play learning algorithm to the consideration of various finite state Mean Field Game settings (finite horizon, $\gamma$-discounted), allowing in particular for the introduction of an additional common noise. We first present a theoretical convergence analysis of the continuous time Fictitious Play process and prove that the induced exploitability decreases at a rate $O(\frac{1}{t})$. Such analysis emphasizes the use of exploitability as a relevant metric for evaluating the convergence towards a Nash equilibrium in the context of Mean Field Games. These theoretical contributions are supported by numerical experiments provided in either model-based or model-free settings. We provide hereby for the first time converging learning dynamics for Mean Field Games in the presence of common noise.

Recent Reinforcement Learning (RL) algorithms making use of Kullback-Leibler (KL) regularization as a core component have shown outstanding performance. Yet, only little is understood theoretically about why KL regularization helps, so far. We study KL regularization within an approximate value iteration scheme and show that it implicitly averages q-values. Leveraging this insight, we provide a very strong performance bound, the very first to combine two desirable aspects: a linear dependency to the horizon (instead of quadratic) and an error propagation term involving an averaging effect of the estimation errors (instead of an accumulation effect). We also study the more general case of an additional entropy regularizer. The resulting abstract scheme encompasses many existing RL algorithms. Some of our assumptions do not hold with neural networks, so we complement this theoretical analysis with an extensive empirical study.

Bootstrapping is a core mechanism in Reinforcement Learning (RL). Most algorithms, based on temporal differences, replace the true value of a transiting state by their current estimate of this value. Yet, another estimate could be leveraged to bootstrap RL: the current policy. Our core contribution stands in a very simple idea: adding the scaled log-policy to the immediate reward. We show that, by slightly modifying Deep Q-Network (DQN) in that way provides an agent that is competitive with the state-of-the-art Rainbow on Atari games, without making use of distributional RL, n-step returns or prioritized replay. To demonstrate the versatility of this idea, we also use it together with an Implicit Quantile Network (IQN). The resulting agent outperforms Rainbow on Atari, installing a new State of the Art with very little modifications to the original algorithm. To add to this empirical study, we provide strong theoretical insights on what happens under the hood -- implicit Kullback-Leibler regularization and increase of the action-gap.

Many recent successful (deep) reinforcement learning algorithms make use of regularization, generally based on entropy or Kullback-Leibler divergence. We propose a general theory of regularized Markov Decision Processes that generalizes these approaches in two directions: we consider a larger class of regularizers, and we consider the general modified policy iteration approach, encompassing both policy iteration and value iteration. The core building blocks of this theory are a notion of regularized Bellman operator and the Legendre-Fenchel transform, a classical tool of convex optimization. This approach allows for error propagation analyses of general algorithmic schemes of which (possibly variants of) classical algorithms such as Trust Region Policy Optimization, Soft Q-learning, Stochastic Actor Critic or Dynamic Policy Programming are special cases. This also draws connections to proximal convex optimization, especially to Mirror Descent.

In this paper, we propose a novel setting for Inverse Reinforcement Learning (IRL), namely "Learning from a Learner" (LfL). As opposed to standard IRL, it does not consist in learning a reward by observing an optimal agent but from observations of another learning (and thus sub-optimal) agent. To do so, we leverage the fact that the observed agent’s policy is assumed to improve over time. The ultimate goal of this approach is to recover the actual environment’s reward and to allow the observer to outperform the learner. To recover that reward in practice, we propose methods based on the entropy-regularized policy iteration framework. We discuss different approaches to learn solely from trajectories in the state-action space. We demonstrate the genericity of our method by observing agents implementing various reinforcement learning algorithms. Finally, we show that, on both discrete and continuous state/action tasks, the observer’s performance (that optimizes the recovered reward) can surpass those of the observed agent.

A Budgeted Markov Decision Process (BMDP) is an extension of a Markov Decision Process to critical applications requiring safety constraints. It relies on a notion of risk implemented in the shape of an upper bound on a constrains violation signal that -- importantly -- can be modified in real-time. So far, BMDPs could only be solved in the case of finite state spaces with known dynamics. This work extends the state-of-the-art to continuous spaces environments and unknown dynamics. We show that the solution to a BMDP is the fixed point of a novel Budgeted Bellman Optimality operator. This observation allows us to introduce natural extensions of Deep Reinforcement Learning algorithms to address large-scale BMDPs. We validate our approach on two simulated applications: spoken dialogue and autonomous driving.

Deep reinforcement learning (RL) has achieved several high profile successes in difficult decision-making problems. However, these algorithms typically require a huge amount of data before they reach reasonable performance. In fact, their performance during learning can be extremely poor. This may be acceptable for a simulator, but it severely limits the applicability of deep RL to many real-world tasks, where the agent must learn in the real environment. In this paper we study a setting where the agent may access data from previous control of the system. We present an algorithm, Deep Q-learning from Demonstrations (DQfD), that leverages small sets of demonstration data to massively accelerate the learning process even from relatively small amounts of demonstration data and is able to automatically assess the necessary ratio of demonstration data while learning thanks to a prioritized replay mechanism. DQfD works by combining temporal difference updates with supervised classification of the demonstrator’s actions. We show that DQfD has better initial performance than Prioritized Dueling Double Deep Q-Networks (PDD DQN) as it starts with better scores on the first million steps on 41 of 42 games and on average it takes PDD DQN 83 million steps to catch up to DQfD’s performance. DQfD learns to out-perform the best demonstration given in 14 of 42 games. In addition, DQfD leverages human demonstrations to achieve state-of-the-art results for 11 games. Finally, we show that DQfD performs better than three related algorithms for incorporating demonstration data into DQN.

Fictitious play is a game theoretic iterative procedure meant to learn an equilibrium in normal form games. However, this algorithm requires that each player has full knowledge of other players’ strategies. Using an architecture inspired by actor-critic algorithms, we build a stochastic approximation of the fictitious play process. This procedure is on-line, decentralized (an agent has no information of others’ strategies and rewards) and applies to multistage games (a generalization of normal form games). In addition, we prove convergence of our method towards a Nash equilibrium in both the cases of zero-sum two-player multistage games and cooperative multistage games. We also provide empirical evidence of the soundness of our approach on the game of Alesia with and without function approximation.

We investigate end-to-end speech-to-text translation on a corpus of audiobooks specifically augmented for this task. Previous works investigated the extreme case where source language transcription is not available during learning nor decoding, but we also study a midway case where source language transcription is available at training time only. In this case, a single model is trained to decode source speech into target text in a single pass. Experimental results show that it is possible to train compact and efficient end-to-end speech translation models in this setup. We also distribute the corpus and hope that our speech translation baseline on this corpus will be challenged in the future.

This paper addresses the problem of learning a Nash equilibrium in $γ$-discounted multiplayer general-sum Markov Games (MGs) in a batch setting. As the number of players increases in MG, the agents may either collaborate or team apart to increase their final rewards. One solution to address this problem is to look for a Nash equilibrium. Although, several techniques were found for the subcase of two-player zero-sum MGs, those techniques fail to find a Nash equilibrium in general-sum Markov Games. In this paper, we introduce a new definition of $ε$-Nash equilibrium in MGs which grasps the strategy’s quality for multiplayer games. We prove that minimizing the norm of two Bellman-like residuals implies to learn such an $ε$-Nash equilibrium. Then, we show that minimizing an empirical estimate of the $L_p$ norm of these Bellman-like residuals allows learning for general-sum games within the batch setting. Finally, we introduce a neural network architecture that successfully learns a Nash equilibrium in generic multiplayer general-sum turn-based MGs.

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.

End-to-end design of dialogue systems has recently become a popular research topic thanks to powerful tools such as encoder-decoder architectures for sequence-to-sequence learning. Yet, most current approaches cast human-machine dialogue management as a supervised learning problem, aiming at predicting the next utterance of a participant given the full history of the dialogue. This vision may fail to correctly render the planning problem inherent to dialogue as well as its contextual and grounded nature. In this paper, we introduce a Deep Reinforcement Learning method to optimize visually grounded task-oriented dialogues, based on the policy gradient algorithm. This approach is tested on the question generation task from the dataset GuessWhat?! containing 120k dialogues and provides encouraging results at solving both the problem of generating natural dialogues and the task of discovering a specific object in a complex picture.

This paper aims at theoretically and empirically comparing two standard optimization criteria for Reinforcement Learning: i) maximization of the mean value and ii) minimization of the Bellman residual. For that purpose, we place ourselves in the framework of policy search algorithms, that are usually designed to maximize the mean value, and derive a method that minimizes the residual $\|T_* v_\pi - v_\pi\|_{1,\nu}$ over policies. A theoretical analysis shows how good this proxy is to policy optimization, and notably that it is better than its value-based counterpart. We also propose experiments on randomly generated generic Markov decision processes, specifically designed for studying the influence of the involved concentrability coefficient. They show that the Bellman residual is generally a bad proxy to policy optimization and that directly maximizing the mean value is much better, despite the current lack of deep theoretical analysis. This might seem obvious, as directly addressing the problem of interest is usually better, but given the prevalence of (projected) Bellman residual minimization in value-based reinforcement learning, we believe that this question is worth to be considered.

It is commonly assumed that language refers to high-level visual concepts while leaving low-level visual processing unaffected. This view dominates the current literature in computational models for language-vision tasks, where visual and linguistic inputs are mostly processed independently before being fused into a single representation. In this paper, we deviate from this classic pipeline and propose to modulate the \emph{entire visual processing} by a linguistic input. Specifically, we introduce Conditional Batch Normalization (CBN) as an efficient mechanism to modulate convolutional feature maps by a linguistic embedding. We apply CBN to a pre-trained Residual Network (ResNet), leading to the MODulatEd ResNet (\MRN) architecture, and show that this significantly improves strong baselines on two visual question answering tasks. Our ablation study confirms that modulating from the early stages of the visual processing is beneficial.

Learning from demonstrations is a paradigm by which an apprentice agent learns a control policy for a dynamic environment by observing demonstrations delivered by an expert agent. It is usually implemented as either imitation learning (IL) or inverse reinforcement learning (IRL) in the literature. On the one hand, IRL is a paradigm relying on the Markov decision processes, where the goal of the apprentice agent is to find a reward function from the expert demonstrations that could explain the expert behavior. On the other hand, IL consists in directly generalizing the expert strategy, observed in the demonstrations, to unvisited states (and it is therefore close to classification, when there is a finite set of possible decisions). While these two visions are often considered as opposite to each other, the purpose of this paper is to exhibit a formal link between these approaches from which new algorithms can be derived. We show that IL and IRL can be redefined in a way that they are equivalent, in the sense that there exists an explicit bijective operator (namely, the inverse optimal Bellman operator) between their respective spaces of solutions. To do so, we introduce the set-policy framework that creates a clear link between the IL and the IRL. As a result, the IL and IRL solutions making the best of both worlds are obtained. In addition, it is a unifying framework from which existing IL and IRL algorithms can be derived and which opens the way for the IL methods able to deal with the environment's dynamics. Finally, the IRL algorithms derived from the set-policy framework are compared with the algorithms belonging to the more common trajectory-matching family. Experiments demonstrate that the set-policy-based algorithms outperform both the standard IRL and IL ones and result in more robust solutions.

The main contribution of this paper consists in extending several non-stationary Reinforcement Learning (RL) algorithms and their theoretical guarantees to the case of γ-discounted zero-sum Markov Games (MGs). As in the case of Markov Decision Processes (MDPs), non-stationary algorithms are shown to exhibit better performance bounds compared to their stationary counterparts. The obtained bounds are generically composed of three terms: 1) a dependency on γ(discount factor), 2) a concentrability coefficient and 3) a propagation error term. This error, depending on the algorithm, can be caused by a regression step, a policy evaluation step or a best-response evaluation step. As a second contribution, we empirically demonstrate, on generic MGs (called Garnets), that non-stationary algorithms outperform their stationary counterparts. In addition, it is shown that their performance mostly depends on the nature of the propagation error. Indeed, algorithms where the error is due to the evaluation of a best-response are penalized (even if they exhibit better concentrability coefficients and dependencies on γ) compared to those suffering from a regression error.

Probabilitic Finite Automata (PFA) are generative graphical models that define distributions with latent variables over finite sequences of symbols, a.k.a. stochastic languages. Traditionally, unsupervised learning of PFA is performed through algorithms that iteratively improves the likelihood like the Expectation-Maximization (EM) algorithm. Recently, learning algorithms based on the so-called Method of Moments (MoM) have been proposed as a much faster alternative that comes with PAC-style guarantees. However, these algorithms do not ensure the learnt automata to model a proper distribution, limiting their applicability and preventing them to serve as an initialization to iterative algorithms. In this paper, we propose a new MoM-based algorithm with PAC-style guarantees that learns automata defining proper distributions. We assess its performances on synthetic problems from the PAutomaC challenge and real datasets extracted from Wikipedia against previous MoM-based algorithms and EM algorithm.

This paper reports theoretical and empirical investigations on the use of quasi-Newton methods to minimize the Optimal Bellman Residual (OBR) of zero-sum two-player Markov Games. First, it reveals that state-of-the-art algorithms can be derived by the direct application of Newton’s method to different norms of the OBR. More precisely, when applied to the norm of the OBR, Newton’s method results in the Bellman Residual Minimization Policy Iteration (BRMPI) and, when applied to the norm of the Projected OBR (POBR), it results into the standard Least Squares Policy Iteration (LSPI) algorithm. Consequently, new algorithms are proposed, making use of quasi-Newton methods to minimize the OBR and the POBR so as to take benefit of enhanced empirical performances at low cost. Indeed, using a quasi-Newton method approach introduces slight modifications in term of coding of LSPI and BRMPI but improves significantly both the stability and the performance of those algorithms. These phenomena are illustrated on an experiment conducted on artificially constructed games called Garnets.

In this paper we introduce a novel model for computer-aided human-human dialogue. In this context, the computer aims at improving the outcome of a human-human task-oriented dialogue by intervening during the course of the interaction. While dialogue state and topic tracking in human-human dialogue have already been studied, few work has been devoted to the sequential part of the problem, where the impact of the system's actions on the future of the conversation is taken into account. This paper addresses this issue by first modelling human-human dialogue as a Markov Reward Process. The task of purposely taking part into the conversation is then optimised within the Linearly Solvable Markov Decision Process framework. Utterances of the Conversational Agent are seen as perturbations in this process, which aim at satisfying the user's long-term goals while keeping the conversation natural. Finally, results obtained by simulation suggest that such an approach is suitable for computer-aided human-human dialogue and is a first step towards three-party dialogue.

In this paper, we introduce a novel framework to encode the dynamics of dialogues into a probabilistic graphical model. Traditionally, Hidden Markov Models (HMMs) would be used to address this problem, involving a first step of hand-crafting to build a dialogue model (e.g. defining potential hidden states) followed by applying expectation-maximisation (EM) algorithms to refine it. Recently, an alternative class of algorithms based on the Method of Moments (MoM) has proven successful in avoiding issues of the EM-like algorithms such as convergence towards local optima, tractability issues, initialization issues or the lack of theoretical guarantees. In this work, we show that dialogues may be modeled by SP-RFA, a class of graphical models efficiently learnable within the MoM and directly usable in planning algorithms (such as reinforcement learning). Experiments are led on the Ubuntu corpus and dialogues are considered as sequences of dialogue acts, represented by their Latent Dirichlet Allocation (LDA) and Latent Semantic Analysis (LSA). We show that a MoM-based algorithm can learn a compact model of sequences of such acts.

Probabilistic Finite Automaton (PFA), Probabilistic Finite State Transducers (PFST) and Hidden Markov Models (HMM) are widely used in Automatic Speech Recognition (ASR), Text-to-Speech (TTS) systems and Part Of Speech (POS) tagging for language modeling. Traditionally, unsupervised learning of these latent variable models is done by Expectation-Maximization (EM)-like algorithms, as the Baum-Welch algorithm. In a recent alternative line of work, learning algorithms based on spectral properties of some low order moments matrices or tensors were proposed. In comparison to EM, they are orders of magnitude faster and come with theoretical convergence guarantees. However, returned models are not ensured to compute proper distributions. They often return negative values that do not sum to one, limiting their applicability and preventing them to serve as an initialization to EM-like algorithms. In this paper, we propose a new spectral algorithm able to learn a large range of models constrained to return proper distributions. We assess its performances on synthetic problems from the PAutomaC challenge and real datasets extracted from Wikipedia. Experiments show that it outperforms previous spectral approaches as well as the Baum-Welch algorithm with random restarts, in addition to serve as an efficient initialization step to EM-like algorithms.

This paper provides an analysis of error propagation in Approximate Dynamic Programming applied to zero-sum two-player Stochastic Games. We provide a novel and unified error propagation analysis in L_p-norm of three well-known algorithms adapted to Stochastic Games (namely Approximate Value Iteration, Approximate Policy Iteration and Approximate Generalized Policy Iteration). We show that we can achieve a stationary policy which is \frac2γ(1 - γ)^2 ε+ \frac1(1 - γ)^2ε’-optimal, where εis the value function approximation error and ε’ is the approximate greedy operator error. In addition, we provide a practical algorithm (AGPI-Q) to solve infinite horizon γ-discounted two-player zero-sum stochastic games in a batch setting. It is an extension of the Fitted-Q algorithm (which solves Markov Decisions Processes in a batch setting) and can be non-parametric. Finally, we demonstrate experimentally the performance of AGPI-Q on a simultaneous two-player game, namely Alesia.

In this paper, an original framework to model human-machine spoken dialogues is proposed to deal with co-adaptation between users and Spoken Dialogue Systems in non-cooperative tasks.The conversation is modeled as a Stochastic Game: both the user and the system have their own preferences but have to come up with an agreement to solve a non-cooperative task.They are jointly trained so the Dialogue Manager learns the optimal strategy against the best possible user.Results obtained by simulation show that non-trivial strategies are learned and that this framework is suitable for dialogue modeling.

In this paper, we propose to reformulate the active learning problem occurring in classification as a sequential decision making problem. We particularly focus on the problem of dynamically allocating a fixed budget of samples. This raises the problem of the trade off between exploration and exploitation which is traditionally addressed in the framework of the multi-armed bandits theory. Based on previous work on bandit theory applied to active learning for regression, we introduce four novel algorithms for solving the online allocation of the budget in a classification problem. Experiments on a generic classification problem demonstrate that these new algorithms compare positively to state-of-the-art methods.

Predictive State Representations (PSRs) are dynamical systems models that keep track of the system's state using predictions of future observations. In contrast to other models of dynamical systems, such as partially observable Markov decision processes, PSRs produces more compact models and can be consistently learned using statistics of the execution trace and spectral decomposition. In this paper we make a connection between rank minimization problems and learning PSRs. This allows us to derive a new algorithm based on nuclear norm minimization. In addition to estimate automatically the dimension of the system, our algorithm compares favorably with the state of art on randomly generated realistic problems of different sizes.

The automatic prediction of the quality of a dialogue is useful to keep track of a spoken dialogue system's performance and, if necessary, adapt its behaviour. Classifiers and regression models have been suggested to make this prediction. The parameters of these models are learnt from a corpus of dialogues evaluated by users or experts. In this paper, we propose to model this task as an ordinal regression problem. We apply support vector machines for ordinal regression on a corpus of dialogues where each system-user exchange was given a rate on a scale of 1 to 5 by experts. Compared to previous models proposed in the literature, the ordinal regression predictor has significantly better results according to the following evaluation metrics: Cohen's agreement rate with experts ratings, Spearman's rank correlation coefficient, and Euclidean and Manhattan errors.

International audience

Modelling time series is quite a difficult task. The last recent years, reservoir computing approaches have been proven very efficient for such problems. Indeed, thanks to recurrence in the connections between neurons, this approach is a powerful tool to catch and model time dependencies between samples. Yet, the prediction quality often depends on the trade-off between the number of neurons in the reservoir and the amount of training data. Supposedly, the larger the number of neurons, the richer the reservoir of dynamics. However, the risk of overfitting problem appears. Conversely, the lower the number of neurons is, the lower the risk of overfitting problem is but also the poorer the reservoir of dynamics is. We consider here the combination of an echo state network with a projection method to benefit from the advantages of the reservoir computing approach without needing to pay attention to overfitting problems due to a lack of training data.

Reinforcement learning (RL) is a machine learning answer to the optimal control problem. It consists of learning an optimal control policy through interactions with the system to be controlled, the quality of this policy being quantified by the so-called value function. A recurrent subtopic of RL concerns computing an approximation of this value function when the system is too large for an exact representation. This survey reviews state-of-the-art methods for (parametric) value function approximation by grouping them into three main categories: bootstrapping, residual, and projected fixed-point approaches. Related algorithms are derived by considering one of the associated cost functions and a specific minimization method, generally a stochastic gradient descent or a recursive least-squares approach.

Spoken Dialogue Systems (SDS) are natural language interfaces for human-computer interaction. User adaptive dialogue management strategies are essential to sustain the naturalness of interaction. In recent years data-driven methods for dialogue optimization have evolved to be a state of art approach. However these methods need vast amounts of corpora for dialogue optimization. In order to cope with the data requirement of these methods, but also to evaluate the dialogue strategies, user simulations are built. Dialogue corpora used to build user simulation are often not annotated in user's perspective and thus can only simulate some generic user behavior, perhaps not representative of any user. This paper aims at clustering dialogue corpora into various groups based on user behaviors observed in the form of full dialogues.

Reinforcement learning (RL) is now part of the state of the art in the domain of spoken dialogue systems (SDS) optimisation. Most performant RL methods, such as those based on Gaussian Processes, require to test small changes in the policy to assess them as improvements or degradations. This process is called on policy learning. Nevertheless, it can result in system behaviours that are not acceptable by users. Learning algorithms should ideally infer an optimal strategy by observing interactions generated by a non-optimal but acceptable strategy, that is learning off-policy. Such methods usually fail to scale up and are thus not suited for real-world systems. In this contribution, a sample-efficient, online and off-policy RL algorithm is proposed to learn an optimal policy. This algorithm is combined to a compact non-linear value function representation (namely a multi-layers perceptron) enabling to handle large scale systems.

This paper adresses the inverse reinforcement learning (IRL) problem, that is inferring a reward for which a demonstrated expert behavior is optimal. We introduce a new algorithm, SCIRL, whose principle is to use the so-called feature expectation of the expert as the parameterization of the score function of a multi-class classifier. This approach produces a reward function for which the expert policy is provably near-optimal. Contrary to most of existing IRL algorithms, SCIRL does not require solving the direct RL problem. Moreover, with an appropriate heuristic, it can succeed with only trajectories sampled according to the expert behavior. This is illustrated on a car driving simulator.

Reinforcement learning (RL) is a machine learning answer to the optimal control problem. It consists of learning an optimal control policy through interactions with the system to be controlled, the quality of this policy being quantified by the so-called value function. An important RL subtopic is to approximate this function when the system is too large for an exact representation. This survey reviews and unifies state of the art methods for parametric value function approximation by grouping them into three main categories: bootstrapping, residuals and projected fixed-point approaches. Related algorithms are derived by considering one of the associated cost functions and a specific way to minimize it, almost always a stochastic gradient descent or a recursive least-squares approach.

Spoken Dialogue Systems (SDS) are man-machine interfaces which use natural language as the medium of interaction. Dialogue corpora collection for the purpose of training and evaluating dialogue systems is an expensive process. User simulators aim at simulating human users in order to generate synthetic data. Existing methods for user simulation mainly focus on generating data with the same statistical consistency as in some reference dialogue corpus. This paper outlines a novel approach for user simulation based on Inverse Reinforcement Learning (IRL). The task of building the user simulator is perceived as a task of imitation learning.

The optimization of dialogue policies using reinforcement learning (RL) is now an accepted part of the state of the art in spoken dialogue systems (SDS). Yet, it is still the case that the commonly used training algorithms for SDS require a large number of dialogues and hence most systems still rely on artificial data generated by a user simulator. Optimization is therefore performed off-line before releasing the system to real users. Gaussian Processes (GP) for RL have recently been applied to dialogue systems. One advantage of GP is that they compute an explicit measure of uncertainty in the value function estimates computed during learning. In this paper, a class of novel learning strategies is described which use uncertainty to control exploration on-line. Comparisons between several exploration schemes show that significant improvements to learning speed can be obtained and that rapid and safe online optimisation is possible, even on a complex task.

Electrocardiogram (ECG) is required during Magnetic Resonance Imaging (MRI) for two reasons, patient monitoring and MRI sequence synchronization for cardiovascular imaging. The MRI environment severely distorts ECG signals. The Magnetic Field Gradients (MFG) especially induce artifacts, which make ECG analysis during MRI acquisition challenging. Specific signal processing is thus required. An MFG artifact modeling has been proposed for their suppression. However the resulting techniques do not take the ECG signals into account during the model parameter estimation. Recently, ECG denoising based on an artificial ECG model and nonlinear Bayesian filtering has been presented. In this paper, a new MFG artifact suppression method based on nonlinear Bayesian filtering and the unification of the ECG and MFG models is proposed. This new approach enables accurate patient monitoring and outperforms state-of-the-art methods in terms of both QRS detection quality and signal to noise ratio.

International audience

International audience

Because reinforcement learning suffers from a lack of scalability, online value (and Q-) function approximation has received increasing interest this last decade. This contribution introduces a novel approximation scheme, namely the Kalman Temporal Differences (KTD) framework, that exhibits the following features: sample-efficiency, non-linear approximation, non-stationarity handling and uncertainty management. A first KTD-based algorithm is provided for deterministic Markov Decision Processes (MDP) which produces biased estimates in the case of stochastic transitions. Than the eXtended KTD framework (XKTD), solving stochastic MDP, is described. Convergence is analyzed for special cases for both deterministic and stochastic transitions. Related algorithms are experimented on classical benchmarks. They compare favorably to the state of the art while exhibiting the announced features.

This paper deals with value function and Q-function approximation in deterministic Markovian decision processes. A general statistical framework based on the Kalman filtering paradigm is introduced. Its principle is to adopt a parametric representation of the value function, to model the associated parameter vector as a random variable and to minimize the mean-squared error of the parameters conditioned on past observed transitions. From this general framework, which will be calledKalmanTemporalDifferences(KTD), and using an approximation scheme called the unscented transform, a family of algorithms is derived, namely KTD-V, KTD-SARSA and KTD-Q, which aim respectively at estimating the value function of a given policy, the Q-function of a given policy and the optimal Q-function. The proposed approach holds for linear and nonlinear parameterization. This framework is discussed and potential advantages and shortcomings are highlighted.

Automatic electrocardiogram (ECG) analysis, especially QRS detection, is still a challenging task. This is even more the case when ECG is acquired during magnetic resonance (MR) examination. The MR environment highly distorts ECG, with Hall effect, due to the important static magnetic field, and artifacts, caused by fast switching magnetic field gradients. Detection of QRS complexes is then affected. In this paper, a new specific MR QRS detector is presented. This method is based on the modulus maximum lines and on the Lipschitz exponent estimation they offer. The use of this regularity characterization enables to distinguish between QRS complexes and MR artifacts. This detector outperforms existing algorithms with almost 99% sensitivity and positive prediction value.

In dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), segmentation of internal kidney structures is essential for functional evaluation. Manual morphological segmentation of cortex, medulla and cavities remains difficult and time-consuming especially because the different renal compartments are hard to distinguish on a single image. We propose to test a semi-automated method to segment internal kidney structures from a DCE-MRI registered sequence. As the temporal intensity evolution is different in each of the three kidney compartments, pixels are sorted according to their time-intensity curves using a k-means partitioning algorithm. No ground truth is available to evaluate resulting segmentations so a manual segmentation by a radiologist is chosen as a reference. We first evaluate some similarity criteria between the functional segmentations and this reference. The same measures are then computed between another manual segmentation and the reference. Results are similar for the two types of comparisons.

Cardiac magnetic resonance imaging (MRI) is very challenging due to the perpetual heart movements. This movement is pseudo-periodic and implies several issues for image acquisition. Building a single image requires several shots, done on a specific timing of the cardiac cycle. Nowadays the heart rate is estimated before imaging and then it is assumed not to evolve during acquisition. Additionally, in order to remove motion artifacts, the patients are asked to perform breath- holds. Unfortunately, while performing a breath-hold, the heart rate is changing in a significant way. To address this problem in the framework of clinical applications, we propose a simple method to predict the out-coming RR interval, in order to compute adapted MR parameters. The RR interval is the time separating two consecutive R waves and corresponds to one cardiac cycle. Due to its simplicity this method is clinically applicable. The prediction is performed by modeling the heart rate variation as a linear combination of different sensors (here, respiratory sensor and amplitude of R wave). We evaluated this method on five healthy subjects in a clinical setup.

Machine learning methods such as reinforcement learning applied to dialogue strategy optimization has become a leading subject of researches since the mid 90's. Indeed, the great variability of factors to take into account makes the design of a spoken dialogue system a tailoring task and reusability of previous work is very difficult. Yet, techniques such as reinforcement learning are very demanding in training data while obtaining a substantial amount of data in the particular case of spoken dialogues is time-consuming and therefore expansive. In order to expand existing data sets, dialogue simulation techniques are becoming a standard solution. In this paper, we present a user model for realistic spoken dialogue simulation and a method for using this model so as to simulate the grounding process. This allows including grounding subdialogues as actions in the reinforcement learning process and learning adapted strategy.

During the last decade, research in the field of Spoken Dialogue Systems (SDS) has experienced increasing growth. However, the design and optimization of SDS is not only about combin- ing speech and language processing systems such as Automatic Speech Recognition (ASR), parsers, Natural Language Gener- ation (NLG), and Text-to-Speech (TTS) synthesis systems. It also requires the development of dialogue strategies taking at least into account the performances of these subsystems (and others), the nature of the task (e.g. form filling, tutoring, robot control, or database search/browsing), and the user's behaviour (e.g. cooperativeness, expertise). Due to the great variability of these factors, reuse of previous hand-crafted designs is also made very difficult. For these reasons, statistical machine learn- ing (ML) methods applied to automatic SDS optimization have been a leading research area for the last few years. In this paper, we provide a short review of the field and of recent advances.

In this paper, we propose to add a model for NLU-related error generation in a modular environment for computer-based simulation of man-machine spoken dialogs. This model is jointly designed with a user model. Both of them are based on the same underlying Bayesian network used with different parameters in such a way that it can generate a consistent user behavior, according to a goal and the interaction history, and been used as a concept classifier. The proposed simulation environment was used to train a reinforcement-learning algorithm on a simple form-filling task and the results of this experiment show that the addition of the NLU model helps pointing out problematic situations that may occur because of misunderstandings and modifying the dialog strategy accordingly

Because of the great variability of factors to take into account, designing a spoken dialogue system is still a tailoring task. Rapid design and reusability of previous work is made very difficult. For these reasons, the application of machine learning methods to dialogue strategy optimization has become a leading subject of researches this last decade. Yet, techniques such as reinforcement learning are very demanding in training data while obtaining a substantial amount of data in the particular case of spoken dialogues is time-consuming and therefore expansive. In order to expand existing data sets, dialogue simulation techniques are becoming a standard solution. In this paper we describe a user modeling technique for realistic simulation of man-machine goal-directed spoken dialogues. This model, based on a stochastic description of man-machine communication, unlike previously proposed models, is consistent along the interaction according to its history and a predefined user goal

The design of Spoken Dialog Systems cannot be considered as the simple combination of speech processing technologies. Indeed, speech-based interface design has been an expert job for a long time. It necessitates good skills in speech technologies and low-level programming. Moreover, rapid development and reusability of previously designed systems remains uneasy. This makes optimality and objective evaluation of design very difficult. The design process is therefore a cyclic process composed of prototype releases, user satisfaction surveys, bug reports and refinements. It is well known that human intervention for testing is time-consuming and above all very expensive. This is one of the reasons for the recent interest in dialog simulation for evaluation as well as for design automation and optimization. In this paper we expose a probabilistic framework for a realistic simulation of spoken dialogs in which the major components of a dialog system are modeled and parameterized thanks to independent data or expert knowledge. Especially, an Automatic Speech Recognition (ASR) system model and a User Model (UM) have been developed. The ASR model, based on articulatory similarities in language models, provides task-adaptive performance prediction and Confidence Level (CL) distribution estimation. The user model relies on the Bayesian Networks (BN) paradigm and is used both for user behavior modeling and Natural Language Understanding (NLU) modeling. The complete simulation framework has been used to train a reinforcement-learning agent on two different tasks. These experiments helped to point out several potentially problematic dialog scenarios.

Speech enabled interfaces and spoken dialog systems are mostly based on statistical speech and language processing modules. Their behavior is therefore not deterministic and hardly predictable. This makes the simulation and the optimization of such systems performances difficult, as well as the reuse of previous work to build new systems. In the aim of a partially automated optimization of such systems, this paper presents a formalism attempt for the description of man-machine spoken communication in the framework of spoken dialog systems. This formalization is partly based on a probabilistic description of the information processing occurring in each module composing a spoken dialog system but also on a stochastic user modeling. Eventually, some possible applications of this theoretic framework are proposed

Speech enabled interfaces are nowadays becoming ubiquitous. The most advanced ones rely on probabilistic pattern matching systems and especially on automatic speech recognition systems. Because of their statistical nature, performances of such systems never reach one hundred percent of correct recognition results. Performances are linked to environmental noise and to intra- and inter-speaker variability of course, but also to the acoustical similarities inside the vocabulary of allowed speech entries, which is usually contextual in the case of man-machine dialogue systems. A good dialogue strategy should therefore dynamically handle the potentiality of recognition errors. In this paper, we compare different methods to model ASR systems in the framework of automatic dialogue strategy optimization and we especially emphasize on a contextdependent ASR modeling method. 1.

Due to recent progresses in the field of speech and natural language processing, spoken dialogue systems are becoming more and more common. Nevertheless, the design of complete dialogue systems remains uneasy. On the one hand, developing such a system involves defining a dialogue strategy. Though automatic learning of dialogue strategies has been introduced in several researches, it stays hard to use in practice. On the other hand, system design implies some coding skills and is still a job for specialists despite the emergence of the voiceXML language. In this paper, we describe a graphical interface dedicated to ease the development of dialogue systems. The user of the interface may be helped along his design thanks to automatically learned strategies.

Though the field of spoken dialogue systems has developed quickly in the last decade, rapid design of dialogue strategies remains uneasy. Several approaches to the problem of automatic strategy learning have been proposed and aie use of Reinforcement Learning introduced by Levin and Pieraccini is becoming part of the state of the art in this area. However, the quality of the strategy learned by the system depends on the definition of the optimization criterion and on the accuracy of aie environment model. In this paper, we propose to bring a model of an ASR system in the simulated environment in order to enhance the learned strategy. To do so, we introduced recognition error rates and confidence levels produced by ASR systems in the optimization criterion.