Knowledge distillation (KD) has gained a lot of attention in the field of model compression for edge devices thanks to its effectiveness in compressing large powerful networks into smaller lower-capacity models. Online distillation, in which both the teacher and the student are learning collaboratively, has also gained much interest due to its ability to improve on the performance of the networks involved. The Kullback-Leibler (KL) divergence ensures the proper knowledge transfer between the teacher and student. However, most online KD techniques present some bottlenecks under the network capacity gap. By cooperatively and simultaneously training, the models the KL distance becomes incapable of properly minimizing the teacher's and student's distributions. Alongside accuracy, critical edge device applications are in need of well-calibrated compact networks. Confidence calibration provides a sensible way of getting trustworthy predictions. We propose BD-KD: Balancing of Divergences for online Knowledge Distillation. We show that adaptively balancing between the reverse and forward divergences shifts the focus of the training strategy to the compact student network without limiting the teacher network's learning process. We demonstrate that, by performing this balancing design at the level of the student distillation loss, we improve upon both performance accuracy and calibration of the compact student network. We conducted extensive experiments using a variety of network architectures and show improvements on multiple datasets including CIFAR-10, CIFAR-100, Tiny-ImageNet, and ImageNet. We illustrate the effectiveness of our approach through comprehensive comparisons and ablations with current state-of-the-art online and offline KD techniques.

Fine-tuning a Pre-trained Language Model (PLM) on a specific downstream task has been a well-known paradigm in Natural Language Processing. However, with the ever-growing size of PLMs, training the entire model on several downstream tasks becomes very expensive and resource-hungry. Recently, different Parameter Efficient Tuning (PET) techniques are proposed to improve the efficiency of fine-tuning PLMs. One popular category of PET methods is the low-rank adaptation methods which insert learnable truncated SVD modules into the original model either sequentially or in parallel. However, low-rank decomposition suffers from limited representation power. In this work, we address this problem using the Kronecker product instead of the low-rank representation. We introduce KronA, a Kronecker product-based adapter module for efficient fine-tuning of Transformer-based PLMs. We apply the proposed methods for fine-tuning T5 on the GLUE benchmark to show that incorporating the Kronecker-based modules can outperform state-of-the-art PET methods.

视觉搜索是指在视觉显示中的一组分散注意力对象之间找到目标对象的任务。在本文中，基于对Coco-Search18数据集的独立分析，我们研究了视觉搜索过程中人类参与者在视觉搜索过程中的表现如何受到不同参数的影响，例如目标对象的大小和偏心率。我们还研究参与者错误率与搜索性能之间的相关性。我们的研究表明，发现更大，更古怪的目标的速度更快，固定次数较少。我们的图形代码可公开可用：

知识蒸馏（KD）是压缩边缘设备深层分类模型的有效工具。但是，KD的表现受教师和学生网络之间较大容量差距的影响。最近的方法已诉诸KD的多个教师助手（TA）设置，该设置依次降低了教师模型的大小，以相对弥合这些模型之间的尺寸差距。本文提出了一种称为“知识蒸馏”课程专家选择的新技术，以有效地增强在容量差距问题下对紧凑型学生的学习。该技术建立在以下假设的基础上：学生网络应逐渐使用分层的教学课程来逐步指导，因为它可以从较低（较高的）容量教师网络中更好地学习（硬）数据样本。具体而言，我们的方法是一种基于TA的逐渐的KD技术，它每个输入图像选择单个教师，该课程是基于通过对图像进行分类的难度驱动的课程的。在这项工作中，我们凭经验验证了我们的假设，并对CIFAR-10，CIFAR-100，CINIC-10和Imagenet数据集进行了严格的实验，并在类似VGG的模型，Resnets和WideresNets架构上显示出提高的准确性。

基于BERT的微调模型在内存，计算和时间上是资源密集的。尽管许多先前的工作旨在通过压缩技术（例如修剪）提高推论效率，但这些作品并未明确解决培训对下游任务的计算挑战。我们介绍了学习者模块和启动，新颖的方法，以利用预训练的语言模型的过度参数化，以获得收敛速度和资源利用率的好处。学习者模块通过微调参数的微调来导航1）有效训练的双结合，以及2）通过确保快速收敛和高度度量得分有效训练。我们在Distilbert上的结果表明，学习者在与基础方面的表现或超过基线。学习者训练7倍的参数比胶水上的最新方法少。在可乐方面，学习者快速调整20％，并且资源利用率显着降低。

我们提出了一种用于图像显着性预测的新方法，群集显着性预测。该方法根据其个人特征和已知的显着图将个体分为群集，并为每个群集生成单独的图像显着模型。我们在个性化显着图的公共数据集上测试了我们的方法，对个人特征因素的重要性各不相同，并观察了对集群的影响。对于每个群集，我们使用图像到图像翻译方法（主要是Pix2Pix模型）将通用显着性图转换为该群集的显着性图。我们尝试了三种最先进的普遍显着性预测方法，即Deepgaze II，ML-Net和Salgan，并看到它们对结果的影响。我们表明，我们的群集显着性预测技术优于最先进的普遍显着性预测模型。我们还通过使用通过受试者相似性聚类算法和两种基线方法比较聚类显着性预测的结果来证明聚类方法的有效性。我们提出了一种方法，将新朋友分配给最合适的集群，基于他们的个人功能和任何已知的显着图。在我们的实验中，我们看到这种将新人分配给群集的方法平均选择了具有更高显着性得分的群集。

视觉硬注意模型主动地选择并观察图像中的一系列子区域以进行预测。大多数难以注意的模型通过首先分析完整的图像来确定关注的地区。然而，可以是最初不可用的整个图像的情况，而是通过一系列部分观测逐渐感测。在本文中，我们设计了一种用于分类这种依次观察的场景的高效难以注意的模型。呈现的模型从未完全观察图像。为了在部分可观察性下选择信息区域，该模型使用贝叶斯最优实验设计。首先，它基于已经观察到的地区合成了不观察区域的特征。然后，应该使用预测的特征来估计所达到的预期信息增益（EIG），应该应该参加各种区域。最后，该模型参加了上述EIG的位置上的实际内容。该模型使用a）复制特征聚合器来维持复制状态，b）线性分类器来预测类标签，c）一个部分变化的自动码器来预测未观察区域的特征。我们使用部分VAE中的归一化流量来处理特征合成问题中的多种模式。我们使用可怜的目标培训我们的模型，并在五个数据集中测试它。当两者都看过几种瞥见时，我们的模型比基线模型更高比基线模型提高了2-10％。

As language models (LMs) scale, they develop many novel behaviors, good and bad, exacerbating the need to evaluate how they behave. Prior work creates evaluations with crowdwork (which is time-consuming and expensive) or existing data sources (which are not always available). Here, we automatically generate evaluations with LMs. We explore approaches with varying amounts of human effort, from instructing LMs to write yes/no questions to making complex Winogender schemas with multiple stages of LM-based generation and filtering. Crowdworkers rate the examples as highly relevant and agree with 90-100% of labels, sometimes more so than corresponding human-written datasets. We generate 154 datasets and discover new cases of inverse scaling where LMs get worse with size. Larger LMs repeat back a dialog user's preferred answer ("sycophancy") and express greater desire to pursue concerning goals like resource acquisition and goal preservation. We also find some of the first examples of inverse scaling in RL from Human Feedback (RLHF), where more RLHF makes LMs worse. For example, RLHF makes LMs express stronger political views (on gun rights and immigration) and a greater desire to avoid shut down. Overall, LM-written evaluations are high-quality and let us quickly discover many novel LM behaviors.

Developments in autonomous vehicles (AVs) are rapidly advancing and will in the next 20 years become a central part to our society. However, especially in the early stages of deployment, there is expected to be incidents involving AVs. In the event of AV incidents, decisions will need to be made that require ethical decisions, e.g., deciding between colliding into a group of pedestrians or a rigid barrier. For an AV to undertake such ethical decision making and path planning, simulation models of the situation will be required that are used in real-time on-board the AV. These models will enable path planning and ethical decision making to be undertaken based on predetermined collision injury severity levels. In this research, models are developed for the path planning and ethical decision making that predetermine knowledge regarding the possible collision injury severities, i.e., peak deformation of the AV colliding into the rigid barrier or the impact velocity of the AV colliding into a pedestrian. Based on such knowledge and using fuzzy logic, a novel nonlinear weighted utility cost function for the collision injury severity levels is developed. This allows the model-based predicted collision outcomes arising from AV peak deformation and AV-pedestrian impact velocity to be examined separately via weighted utility cost functions with a common structure. The general form of the weighted utility cost function exploits a fuzzy sets approach, thus allowing common utility costs from the two separate utility cost functions to be meaningfully compared. A decision-making algorithm, which makes use of a utilitarian ethical approach, ensures that the AV will always steer onto the path which represents the lowest injury severity level, hence utility cost to society.

Electronic Health Records (EHRs) hold detailed longitudinal information about each patient's health status and general clinical history, a large portion of which is stored within the unstructured text. Temporal modelling of this medical history, which considers the sequence of events, can be used to forecast and simulate future events, estimate risk, suggest alternative diagnoses or forecast complications. While most prediction approaches use mainly structured data or a subset of single-domain forecasts and outcomes, we processed the entire free-text portion of EHRs for longitudinal modelling. We present Foresight, a novel GPT3-based pipeline that uses NER+L tools (i.e. MedCAT) to convert document text into structured, coded concepts, followed by providing probabilistic forecasts for future medical events such as disorders, medications, symptoms and interventions. Since large portions of EHR data are in text form, such an approach benefits from a granular and detailed view of a patient while introducing modest additional noise. On tests in two large UK hospitals (King's College Hospital, South London and Maudsley) and the US MIMIC-III dataset precision@10 of 0.80, 0.81 and 0.91 was achieved for forecasting the next biomedical concept. Foresight was also validated on 34 synthetic patient timelines by 5 clinicians and achieved relevancy of 97% for the top forecasted candidate disorder. Foresight can be easily trained and deployed locally as it only requires free-text data (as a minimum). As a generative model, it can simulate follow-on disorders, medications and interventions for as many steps as required. Foresight is a general-purpose model for biomedical concept modelling that can be used for real-world risk estimation, virtual trials and clinical research to study the progression of diseases, simulate interventions and counterfactuals, and for educational purposes.