深度学习一直是近来最具破坏性的技术进步之一。深度学习模型的高性能以高度计算，存储和功率要求为代价。感知到加速和压缩这些模型以提高设备性能的直接需求，我们引入了Deeplite Neutrino，以便对模型的生产优化和Deeplite运行时进行介绍，以在基于ARM的平台上部署超低位量化模型。我们为ARMV7和ARMV8架构实施低级量化内核，可在32位和64位基于ARM的设备上进行部署。通过使用矢量化，并行化和平铺的有效实现，与具有XNNPACK后端的TensorFlow Lite相比，我们在分类和检测模型上分别实现了高达2倍和2.2倍的速度。与ONNX运行时相比，我们还获得了高达5倍和3.2倍的显着加速，分别用于分类和检测模型。

在本文中，我们研究了DNN培训中量化的影响。我们假设重量量化是正则化的一种形式，正则化的量与量化水平（精度）相关。我们通过提供分析研究和经验结果来证实我们的假设。通过将重量量化为重量噪声的一种形式，我们探讨了该噪声在训练时如何通过网络传播。然后，我们表明该噪声的大小与量化水平相关。为了确认我们的分析研究，我们在本文中进行了广泛的实验列表，其中我们表明，在各种数据集中，在各种视觉任务和模型中可以看到量化的正则化效果。基于我们的研究，我们建议8位量化在不同的视觉任务和模型中提供了一种可靠的正则化形式。

神经网络（NNS）对研究和行业进行了很大的影响。然而，随着NNS的准确性增加，它之后的大小是扩展，所需的计算操作数量和能量消耗。资源消费的增加导致NNS减少的采用率和现实世界部署不切实际。因此，需要压缩NNS以使它们可用于更广泛的受众，同时降低其运行时成本。在这项工作中，我们从因果推理的角度来处理这一挑战，我们提出了一个评分机制，以促进NNS的结构灌注。该方法基于在最大熵扰动下测量互信息，顺序地通过NN传播。我们展示了两种数据集和各种NNS尺寸的方法的表现，我们表明我们的方法在挑战条件下实现了竞争性能。

Finding and localizing the conceptual changes in two scenes in terms of the presence or removal of objects in two images belonging to the same scene at different times in special care applications is of great significance. This is mainly due to the fact that addition or removal of important objects for some environments can be harmful. As a result, there is a need to design a program that locates these differences using machine vision. The most important challenge of this problem is the change in lighting conditions and the presence of shadows in the scene. Therefore, the proposed methods must be resistant to these challenges. In this article, a method based on deep convolutional neural networks using transfer learning is introduced, which is trained with an intelligent data synthesis process. The results of this method are tested and presented on the dataset provided for this purpose. It is shown that the presented method is more efficient than other methods and can be used in a variety of real industrial environments.

Transformers have recently gained attention in the computer vision domain due to their ability to model long-range dependencies. However, the self-attention mechanism, which is the core part of the Transformer model, usually suffers from quadratic computational complexity with respect to the number of tokens. Many architectures attempt to reduce model complexity by limiting the self-attention mechanism to local regions or by redesigning the tokenization process. In this paper, we propose DAE-Former, a novel method that seeks to provide an alternative perspective by efficiently designing the self-attention mechanism. More specifically, we reformulate the self-attention mechanism to capture both spatial and channel relations across the whole feature dimension while staying computationally efficient. Furthermore, we redesign the skip connection path by including the cross-attention module to ensure the feature reusability and enhance the localization power. Our method outperforms state-of-the-art methods on multi-organ cardiac and skin lesion segmentation datasets without requiring pre-training weights. The code is publicly available at https://github.com/mindflow-institue/DAEFormer.

The stock market prediction has been a traditional yet complex problem researched within diverse research areas and application domains due to its non-linear, highly volatile and complex nature. Existing surveys on stock market prediction often focus on traditional machine learning methods instead of deep learning methods. Deep learning has dominated many domains, gained much success and popularity in recent years in stock market prediction. This motivates us to provide a structured and comprehensive overview of the research on stock market prediction focusing on deep learning techniques. We present four elaborated subtasks of stock market prediction and propose a novel taxonomy to summarize the state-of-the-art models based on deep neural networks from 2011 to 2022. In addition, we also provide detailed statistics on the datasets and evaluation metrics commonly used in the stock market. Finally, we highlight some open issues and point out several future directions by sharing some new perspectives on stock market prediction.

Existing training criteria in automatic speech recognition(ASR) permit the model to freely explore more than one time alignments between the feature and label sequences. In this paper, we use entropy to measure a model's uncertainty, i.e. how it chooses to distribute the probability mass over the set of allowed alignments. Furthermore, we evaluate the effect of entropy regularization in encouraging the model to distribute the probability mass only on a smaller subset of allowed alignments. Experiments show that entropy regularization enables a much simpler decoding method without sacrificing word error rate, and provides better time alignment quality.

Climate change has increased the intensity, frequency, and duration of extreme weather events and natural disasters across the world. While the increased data on natural disasters improves the scope of machine learning (ML) in this field, progress is relatively slow. One bottleneck is the lack of benchmark datasets that would allow ML researchers to quantify their progress against a standard metric. The objective of this short paper is to explore the state of benchmark datasets for ML tasks related to natural disasters, categorizing them according to the disaster management cycle. We compile a list of existing benchmark datasets introduced in the past five years. We propose a web platform - NADBenchmarks - where researchers can search for benchmark datasets for natural disasters, and we develop a preliminary version of such a platform using our compiled list. This paper is intended to aid researchers in finding benchmark datasets to train their ML models on, and provide general directions for topics where they can contribute new benchmark datasets.

Recent pre-trained language models have shown promising capabilities in generating fluent and realistic natural language text. However, generating multi-sentence text with global content planning has been a long-existing research question. Current approaches for controlled text generation can hardly address this issue, as they usually condition on single known control attributes. In this study, we propose a low-cost yet effective framework which explicitly models the global content plan of the generated text. Specifically, it optimizes the joint distribution of the natural language sequence and the global content plan in a plug-and-play manner. We conduct extensive experiments on the well-established Recipe1M+ benchmark. Both automatic and human evaluations verify that our model achieves the state-of-the-art performance on the task of recipe generation

We present a new algorithm to learn a deep neural network model robust against adversarial attacks. Previous algorithms demonstrate an adversarially trained Bayesian Neural Network (BNN) provides improved robustness. We recognize the adversarial learning approach for approximating the multi-modal posterior distribution of a Bayesian model can lead to mode collapse; consequently, the model's achievements in robustness and performance are sub-optimal. Instead, we first propose preventing mode collapse to better approximate the multi-modal posterior distribution. Second, based on the intuition that a robust model should ignore perturbations and only consider the informative content of the input, we conceptualize and formulate an information gain objective to measure and force the information learned from both benign and adversarial training instances to be similar. Importantly. we prove and demonstrate that minimizing the information gain objective allows the adversarial risk to approach the conventional empirical risk. We believe our efforts provide a step toward a basis for a principled method of adversarially training BNNs. Our model demonstrate significantly improved robustness--up to 20%--compared with adversarial training and Adv-BNN under PGD attacks with 0.035 distortion on both CIFAR-10 and STL-10 datasets.