In this paper, we present Tianshou, a highly modularized Python library for deep reinforcement learning (DRL) that uses PyTorch as its backend. Tianshou intends to be research-friendly by providing flexible and reliable infrastructure of DRL algorithms. It supports online offline training with more than 20 classic algorithms through unified interface. To facilitate related research prove Tianshou's reliability, have released benchmark MuJoCo environments, covering eight state-of-the-art...

We develop herein an efficient copper-catalyzed dihydroboration of alkynes with HBpin to synthesize gem-diborylalkanes. This catalytic process features a broad substrate scope, good functional group tolerance, and excellent regioselectivity. Notably, the gram-scale reaction further product derivatizations demonstrate practicality this protocol. Preliminary mechanistic studies indicate that involves hydroboration alkyne vinylboronate intermediate.

We propose to address quadrupedal locomotion tasks using Reinforcement Learning (RL) with a Transformer-based model that learns combine proprioceptive information and high-dimensional depth sensor inputs. While learning-based has made great advances RL, most methods still rely on domain randomization for training blind agents generalize challenging terrains. Our key insight is states only offer contact measurements immediate reaction, whereas an agent equipped visual sensory observations can...

Developing robust vision-guided controllers for quadrupedal robots in complex environments with various obstacles, dynamical surroundings and uneven terrains is very challenging. While Reinforcement Learning (RL) provides a promising paradigm agile locomotion skills vision inputs simulation, it still challenging to deploy the RL policy real world. Our key insight that asynchronous multi-modal observations, caused by different latencies components of robot, create large sim2real gap policy....

Reconstructing a 3D shape from single-view image using deep learning has become increasingly popular recently. Most existing methods only focus on reconstructing the geometry based constraints. The lack of explicit modeling structure relations among parts yields low-quality reconstruction results for structure-rich man-made shapes. In addition, conventional 2D-3D joint embedding architecture image-based often omits specific view information given image, which may lead to degraded and...

Peach (Prunus persica (L.) Batsch) is a typical shallow-rooted fruit plant with high respiratory intensity and oxygen demand, which makes it highly susceptible to oxygen-deficient soil conditions resulting from waterlogging. Rootstock waterlogging resistance essential the performance of cultivated peaches under stress. In comparison Prunus var. (‘Maotao’, M) davidiana (Carr.) C. de Vos (‘Shantao’, S), f. Hossu (‘Hossu’, H) exhibited superior leaf photosynthetic electron transfer efficiency,...

In plants, a family of terpene synthases (TPSs) is responsible for the biosynthesis terpenes and contributes to species-specific diversity volatile organic compounds, which play essential roles in fitness plants. However, little known about TPS gene peach and/or nectarine (Prunus persica L.). this study, we identified 40 PpTPS genes genome v2.0. Although these PpTPSs could be clustered into five classes, they distribute several clusters three chromosomes, share conserved exon-intron...

Supplier network collaborative efficiency evaluation is important content in the transformation and upgrading of intelligent manufacturing enterprises. Aiming at shortcomings existing methods, this paper proposes a new method to evaluate internal members complex supplier based on theory. Based analysis characteristics network, from perspective system, macro divided into multiple multi-level micro subsystems with enterprises as core. In order reasonably quantify collaboration relationship...

Sample efficiency has been one of the major challenges for deep reinforcement learning. Recently, model-based learning proposed to address this challenge by performing planning on imaginary trajectories with a learned world model. However, model may suffer from overfitting training trajectories, and thus value estimation policy search will be pone sucked in an inferior local policy. In paper, we propose novel algorithm, called BrIdging Reality Dream (BIRD). It maximizes mutual information...

As a typical time series, the length of data sequence is critical to accuracy traffic state prediction. In order fully explore causality between data, this study established temporal backtracking and multistep delay model based on recurrent neural networks (RNNs) learn extract long- short-term dependencies data. With real set, coordinate descent algorithm was employed search determine optimal sequence, predictions were performed demonstrate relationship steps prediction accuracies. Besides,...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Retrieval-Augmented Generation (RAG) mitigates the issue of hallucination in Large Language Models (LLMs) by integrating information retrieval techniques. However, tourism domain, since query is usually brief and content database diverse, existing RAG may contain a significant amount irrelevant or contradictory contents after retrieval. To address this challenge, we propose QCG-Rerank model. This model first performs an initial to obtain candidate chunks then enhances semantics extracting...

Two-phase flow instability may occur in nuclear reactor systems, which is often accompanied by periodic fluctuation fluid rate. In this study, bubble rising and coalescence characteristics under inlet pulsation condition are analyzed based on the MPS-MAFL method. To begin with, single behavior was simulated. The simulation results show that shape velocity fluctuate periodically as same Additionally, pairs’ simulated compared with static results. It found time of pairs slightly increased...

We address the problem of safely solving complex bimanual robot manipulation tasks with sparse rewards. Such challenging can be decomposed into sub-tasks that are accomplishable by different robots concurrently or sequentially for better efficiency. While previous reinforcement learning approaches primarily focus on modeling compositionality sub-tasks, two fundamental issues largely ignored particularly when cooperative strategies robots: (i) domination, i.e., one may try to solve a task...

In view of the current situation that maturity enterprise intelligent manufacturing capability is generally low and information asymmetry in upstream downstream supply chain high, taking any demand link as an example, a group initial signals change nonlinearly over time are divided into intrinsic mode functions noise residuals with different data characteristics by means variational modal decomposition (VMD) algorithm. On basis signal denoising reconstruction, support vector machine (SVM)...

In an integrated energy system, due to the characteristics of multi-energy flow in it brings inconvenience overall analysis, and multi-system coupling allow some faults propagate between different systems. The topology network results existence key nodes system. For analysis structure, weak points that affect security system can be found, which provide guidance supplements for subsequent assessments.

A new image translation method is proposed to solve the problems existing in current models, such as detail loss generated images and training instability. In this paper, ResNet added generator network degradation problem process of deep through residual learning. After each convolution layer discriminator network, a spectral normalization make model more stable improve quality image. Experimental results show that, compared with other methods, by has better effect lower FID.

Safety assessment is of great significance in the development and promotion integrated energy system. In order to make traditional safety can be better applied large-scale systems, this paper optimizes process from aspects generation expected fault sets solution multi-energy flow power flow. Implement improvements security process.

Lithium-ion batteries function involves coupled chemomechanical phenomena spanning multiple length scales. An in-depth understanding and quantifying of the battery degradation mechanisms constitutes a frontier challenge that requires multi-scale characterization. Here we present macro-to-nano scale analysis through hierarchy thick NMC811 (LiNi0.8Mn0.1Co0.1O2) electrode for lithium-ion using suite state-of-the-art advanced techniques. Emerging plasma focused ion beam-scanning electron...