Implicit Subgoal Planning with Variational Autoencoders for Long-Horizon Sparse Reward Robotic Tasks

The challenges inherent to long-horizon tasks in robotics persist due to the typical inefficient exploration and sparse rewards in traditional reinforcement learning approaches. To alleviate these challenges, we introduce a novel algorithm, Variational Autoencoder-based Subgoal Inference (VAESI), to accomplish long-horizon tasks through a divide-and-conquer manner. VAESI consists of three components: a Variational Autoencoder (VAE)-based Subgoal Generator, a Hindsight Sampler, and a Value Selector. The VAE-based Subgoal Generator draws inspiration from the human capacity to infer subgoals and reason about the final goal in the context of these subgoals. It is composed of an explicit encoder model, engineered to generate subgoals, and an implicit decoder model, designed to enhance the quality of the generated subgoals by predicting the final goal. Additionally, the Hindsight Sampler selects valid subgoals from an offline dataset to enhance the feasibility of the generated subgoals. The Value Selector utilizes the value function in reinforcement learning to filter the optimal subgoals from subgoal candidates. To validate our method, we conduct several long-horizon tasks in both simulation and the real world, including one locomotion task and three manipulation tasks. The obtained quantitative and qualitative data indicate that our approach achieves promising performance compared to other baseline methods. These experimental results can be seen in the website \url{https://sites.google.com/view/vaesi/home}.

ApproBiVT: Lead ASR Models to Generalize Better Using Approximated Bias-Variance Tradeoff Guided Early Stopping and Checkpoint Averaging

The conventional recipe for Automatic Speech Recognition (ASR) models is to 1) train multiple checkpoints on a training set while relying on a validation set to prevent overfitting using early stopping and 2) average several last checkpoints or that of the lowest validation losses to obtain the final model. In this paper, we rethink and update the early stopping and checkpoint averaging from the perspective of the bias-variance tradeoff. Theoretically, the bias and variance represent the fitness and variability of a model and the tradeoff of them determines the overall generalization error. But, it's impractical to evaluate them precisely. As an alternative, we take the training loss and validation loss as proxies of bias and variance and guide the early stopping and checkpoint averaging using their tradeoff, namely an Approximated Bias-Variance Tradeoff (ApproBiVT). When evaluating with advanced ASR models, our recipe provides 2.5%-3.7% and 3.1%-4.6% CER reduction on the AISHELL-1 and AISHELL-2, respectively.

P-vectors: A Parallel-Coupled TDNN/Transformer Network for Speaker Verification

Typically, the Time-Delay Neural Network (TDNN) and Transformer can serve as a backbone for Speaker Verification (SV). Both of them have advantages and disadvantages from the perspective of global and local feature modeling. How to effectively integrate these two style features is still an open issue. In this paper, we explore a Parallel-coupled TDNN/Transformer Network (p-vectors) to replace the serial hybrid networks. The p-vectors allows TDNN and Transformer to learn the complementary information from each other through Soft Feature Alignment Interaction (SFAI) under the premise of preserving local and global features. Also, p-vectors uses the Spatial Frequency-channel Attention (SFA) to enhance the spatial interdependence modeling for input features. Finally, the outputs of dual branches of p-vectors are combined by Embedding Aggregation Layer (EAL). Experiments show that p-vectors outperforms MACCIF-TDNN and MFA-Conformer with relative improvements of 11.5% and 13.9% in EER on VoxCeleb1-O.

Sequentially Sampled Chunk Conformer for Streaming End-to-End ASR

This paper presents an in-depth study on a Sequentially Sampled Chunk Conformer, SSC-Conformer, for streaming End-to-End (E2E) ASR. The SSC-Conformer first demonstrates the significant performance gains from using the sequentially sampled chunk-wise multi-head self-attention (SSC-MHSA) in the Conformer encoder by allowing efficient cross-chunk interactions while keeping linear complexities. Furthermore, it explores taking advantage of chunked convolution to make use of the chunk-wise future context and integrates with casual convolution in the convolution layers to further reduce CER. We verify the proposed SSC-Conformer on the AISHELL-1 benchmark and experimental results show that a state-of-the-art performance for streaming E2E ASR is achieved with CER 5.33% without LM rescoring. And, owing to its linear complexity, the SSC-Conformer can train with large batch sizes and infer more efficiently.

Sequentially Sampled Chunk Conformer or Streaming End-to-End ASR

This paper presents an in-depth study on a Sequentially Sampled Chunk Conformer, SSC-Conformer, for streaming End-to-End (E2E) ASR. The SSC-Conformer first demonstrates the significant performance gains from using the sequentially sampled chunk-wise multi-head self-attention (SSC-MHSA) in the Conformer encoder by allowing efficient cross-chunk interactions while keeping linear complexities. Furthermore, it explores taking advantage of chunked convolution to make use of the chunk-wise future context and integrates with casual convolution in the convolution layers to further reduce CER. We verify the proposed SSC-Conformer on the AISHELL-1 benchmark and experimental results show that a state-of-the-art performance for streaming E2E ASR is achieved with CER 5.33% without LM rescoring. And, owing to its linear complexity, the SSC-Conformer can train with large batch sizes and infer more efficiently.

A Dual-Arm Collaborative Framework for Dexterous Manipulation in Unstructured Environments with Contrastive Planning

Most object manipulation strategies for robots are based on the assumption that the object is rigid (i.e., with fixed geometry) and the goal's details have been fully specified (e.g., the exact target pose). However, there are many tasks that involve spatial relations in human environments where these conditions may be hard to satisfy, e.g., bending and placing a cable inside an unknown container. To develop advanced robotic manipulation capabilities in unstructured environments that avoid these assumptions, we propose a novel long-horizon framework that exploits contrastive planning in finding promising collaborative actions. Using simulation data collected by random actions, we learn an embedding model in a contrastive manner that encodes the spatio-temporal information from successful experiences, which facilitates the subgoal planning through clustering in the latent space. Based on the keypoint correspondence-based action parameterization, we design a leader-follower control scheme for the collaboration between dual arms. All models of our policy are automatically trained in simulation and can be directly transferred to real-world environments. To validate the proposed framework, we conduct a detailed experimental study on a complex scenario subject to environmental and reachability constraints in both simulation and real environments.

An Embedded Feature Selection Framework for Control

Reducing sensor requirements while keeping optimal control performance is crucial to many industrial control applications to achieve robust, low-cost, and computation-efficient controllers. However, existing feature selection solutions for the typical machine learning domain can hardly be applied in the domain of control with changing dynamics. In this paper, a novel framework, namely the Dual-world embedded Attentive Feature Selection (D-AFS), can efficiently select the most relevant sensors for the system under dynamic control. Rather than the one world used in most Deep Reinforcement Learning (DRL) algorithms, D-AFS has both the real world and its virtual peer with twisted features. By analyzing the DRL's response in two worlds, D-AFS can quantitatively identify respective features' importance towards control. A well-known active flow control problem, cylinder drag reduction, is used for evaluation. Results show that D-AFS successfully finds an optimized five-probes layout with 18.7\% drag reduction than the state-of-the-art solution with 151 probes and 49.2\% reduction than five-probes layout by human experts. We also apply this solution to four OpenAI classical control cases. In all cases, D-AFS achieves the same or better sensor configurations than originally provided solutions. Results highlight, we argued, a new way to achieve efficient and optimal sensor designs for experimental or industrial systems. Our source codes are made publicly available at https://github.com/G-AILab/DAFSFluid.

Shifted Chunk Encoder for Transformer Based Streaming End-to-End ASR

Currently, there are mainly three Transformer encoder based streaming End to End (E2E) Automatic Speech Recognition (ASR) approaches, namely time-restricted methods, chunk-wise methods, and memory based methods. However, all of them have some limitations in aspects of global context modeling, linear computational complexity, and model parallelism. In this work, we aim to build a single model to achieve the benefits of all the three aspects for streaming E2E ASR. Particularly, we propose to use a shifted chunk mechanism instead of the conventional chunk mechanism for streaming Transformer and Conformer. This shifted chunk mechanism can significantly enhance modeling power through allowing chunk self-attention to capture global context across local chunks, while keeping linear computational complexity and parallel trainable. We name the Shifted Chunk Transformer and Conformer as SChunk-Transofromer and SChunk-Conformer, respectively. And we verify their performance on the widely used AISHELL-1 benckmark. Experiments show that the SChunk-Transformer and SChunk-Conformer achieve CER 6.43% and 5.77%, respectively. That surpasses the existing chunk-wise and memory based methods by a large margin, and is competitive even compared with the state-of-the-art time-restricted methods which have quadratic computational complexity.

MACCIF-TDNN: Multi aspect aggregation of channel and context interdependence features in TDNN-based speaker verification

Most of the recent state-of-the-art results for speaker verification are achieved by X-vector and its subsequent variants. In this paper, we propose a new network architecture which aggregates the channel and context interdependence features from multi aspect based on Time Delay Neural Network (TDNN). Firstly, we use the SE-Res2Blocks as in ECAPA-TDNN to explicitly model the channel interdependence to realize adaptive calibration of channel features, and process local context features in a multi-scale way at a more granular level compared with conventional TDNN-based methods. Secondly, we explore to use the encoder structure of Transformer to model the global context interdependence features at an utterance level which can capture better long term temporal characteristics. Before the pooling layer, we aggregate the outputs of SE-Res2Blocks and Transformer encoder to leverage the complementary channel and context interdependence features learned by themself respectively. Finally, instead of performing a single attentive statistics pooling, we also find it beneficial to extend the pooling method in a multi-head way which can discriminate features from multiple aspect. The proposed MACCIF-TDNN architecture can outperform most of the state-of-the-art TDNN-based systems on VoxCeleb1 test sets.