Masked time series modeling has recently gained much attention as a self-supervised representation learning strategy for time series. Inspired by masked image modeling in computer vision, recent works first patchify and partially mask out time series, and then train Transformers to capture the dependencies between patches by predicting masked patches from unmasked patches. However, we argue that capturing such patch dependencies might not be an optimal strategy for time series representation learning; rather, learning to embed patches independently results in better time series representations. Specifically, we propose to use 1) the simple patch reconstruction task, which autoencode each patch without looking at other patches, and 2) the simple patch-wise MLP that embeds each patch independently. In addition, we introduce complementary contrastive learning to hierarchically capture adjacent time series information efficiently. Our proposed method improves time series forecasting and classification performance compared to state-of-the-art Transformer-based models, while it is more efficient in terms of the number of parameters and training/inference time. Code is available at this repository: //github.com/seunghan96/pits.
相關內容
As deep learning models become increasingly bigger and more complex, it is critical to improve model training and inference efficiency. Though a variety of highly optimized libraries and packages (known as DL kernels) have been developed, it is tedious and time-consuming to figure out which kernel to use, where to use, and how to use them correctly. To address this challenge, we propose an Automated Deep learning OPTimization approach called Adopter. We design a Domain-Specific Language (DSL) to represent DL model architectures and leverage this DSL to specify model transformation rules required to integrate a DL kernel into a model. Given the source code of a DL model and the transformation rules for a set of kernels, Adopter first performs inter-procedural analysis to identify and express the model architecture in our DSL. Then, Adopter performs scope analysis and sub-sequence matching to identify locations in the model architecture where the transformation rules can be applied. Finally, Adopter proposes a synthesis-based code transformation method to apply the transformation rule. We curated a benchmark with 199 models from Hugging Face and a diverse set of DL kernels. We found that, compared to a state-of-the-art automated code transformation technique, Adopter helps improve the precision and recall by 3% and 56%, respectively. An in-depth analysis of 9 models revealed that on average, Adopter improved the training speed by 22.7% while decreasing the GPU memory usage by 10.5%.
Equilibrium Propagation (EP) is a biologically plausible local learning algorithm initially developed for convergent recurrent neural networks (RNNs), where weight updates rely solely on the connecting neuron states across two phases. The gradient calculations in EP have been shown to approximate the gradients computed by Backpropagation Through Time (BPTT) when an infinitesimally small nudge factor is used. This property makes EP a powerful candidate for training Spiking Neural Networks (SNNs), which are commonly trained by BPTT. However, in the spiking domain, previous studies on EP have been limited to architectures involving few linear layers. In this work, for the first time we provide a formulation for training convolutional spiking convergent RNNs using EP, bridging the gap between spiking and non-spiking convergent RNNs. We demonstrate that for spiking convergent RNNs, there is a mismatch in the maximum pooling and its inverse operation, leading to inaccurate gradient estimation in EP. Substituting this with average pooling resolves this issue and enables accurate gradient estimation for spiking convergent RNNs. We also highlight the memory efficiency of EP compared to BPTT. In the regime of SNNs trained by EP, our experimental results indicate state-of-the-art performance on the MNIST and FashionMNIST datasets, with test errors of 0.97% and 8.89%, respectively. These results are comparable to those of convergent RNNs and SNNs trained by BPTT. These findings underscore EP as an optimal choice for on-chip training and a biologically-plausible method for computing error gradients.
Domain experts can play a crucial role in guiding data scientists to optimize machine learning models while ensuring contextual relevance for downstream use. However, in current workflows, such collaboration is challenging due to differing expertise, abstract documentation practices, and lack of access and visibility into low-level implementation artifacts. To address these challenges and enable domain expert participation, we introduce CellSync, a collaboration framework comprising (1) a Jupyter Notebook extension that continuously tracks changes to dataframes and model metrics and (2) a Large Language Model powered visualization dashboard that makes those changes interpretable to domain experts. Through CellSync's cell-level dataset visualization with code summaries, domain experts can interactively examine how individual data and modeling operations impact different data segments. The chat features enable data-centric conversations and targeted feedback to data scientists. Our preliminary evaluation shows that CellSync provides transparency and promotes critical discussions about the intents and implications of data operations.
This study explores the application of deep learning technologies in software development processes, particularly in automating code reviews, error prediction, and test generation to enhance code quality and development efficiency. Through a series of empirical studies, experimental groups using deep learning tools and control groups using traditional methods were compared in terms of code error rates and project completion times. The results demonstrated significant improvements in the experimental group, validating the effectiveness of deep learning technologies. The research also discusses potential optimization points, methodologies, and technical challenges of deep learning in software development, as well as how to integrate these technologies into existing software development workflows.
Of all the vector fields surrounding the minima of recurrent learning setups, the gradient field with its exploding and vanishing updates appears a poor choice for optimization, offering little beyond efficient computability. We seek to improve this suboptimal practice in the context of physics simulations, where backpropagating feedback through many unrolled time steps is considered crucial to acquiring temporally coherent behavior. The alternative vector field we propose follows from two principles: physics simulators, unlike neural networks, have a balanced gradient flow, and certain modifications to the backpropagation pass leave the positions of the original minima unchanged. As any modification of backpropagation decouples forward and backward pass, the rotation-free character of the gradient field is lost. Therefore, we discuss the negative implications of using such a rotational vector field for optimization and how to counteract them. Our final procedure is easily implementable via a sequence of gradient stopping and component-wise comparison operations, which do not negatively affect scalability. Our experiments on three control problems show that especially as we increase the complexity of each task, the unbalanced updates from the gradient can no longer provide the precise control signals necessary while our method still solves the tasks. Our code can be found at //github.com/tum-pbs/StableBPTT.
There are many different forms of design knowledge that guide and shape a designer's ability to act and realize potential realities. Methods and schemas are examples of design knowledge commonly used by design researchers and designers alike. In this pictorial, we explore, engage, and describe the role of schemas as tools that can support design researchers in formulating methods to support design action, with our framing of method design specifically focused on ethical design complexity. We present four ways for method designers to engage with schema: 1) Systems to operationalize complex design constructs such as ethical design complexity through an A.E.I.O.YOU schema; 2) Classifiers to map existing methods and identify the possibility for new methods through descriptive semantic differentials; 3) Tools that enable the creation of methods that relate to one or more elements of the schema through creative departures from research to design; and 4) Interactive channels to playfully engage potential and new opportunities through schema interactivity.
Interventions targeting the representation space of language models (LMs) have emerged as an effective means to influence model behavior. Such methods are employed, for example, to eliminate or alter the encoding of demographic information such as gender within the model's representations and, in so doing, create a counterfactual representation. However, because the intervention operates within the representation space, understanding precisely what aspects of the text it modifies poses a challenge. In this paper, we give a method to convert representation counterfactuals into string counterfactuals. We demonstrate that this approach enables us to analyze the linguistic alterations corresponding to a given representation space intervention and to interpret the features utilized to encode a specific concept. Moreover, the resulting counterfactuals can be used to mitigate bias in classification through data augmentation.
Contrastive learning models have achieved great success in unsupervised visual representation learning, which maximize the similarities between feature representations of different views of the same image, while minimize the similarities between feature representations of views of different images. In text summarization, the output summary is a shorter form of the input document and they have similar meanings. In this paper, we propose a contrastive learning model for supervised abstractive text summarization, where we view a document, its gold summary and its model generated summaries as different views of the same mean representation and maximize the similarities between them during training. We improve over a strong sequence-to-sequence text generation model (i.e., BART) on three different summarization datasets. Human evaluation also shows that our model achieves better faithfulness ratings compared to its counterpart without contrastive objectives.
We advocate the use of implicit fields for learning generative models of shapes and introduce an implicit field decoder for shape generation, aimed at improving the visual quality of the generated shapes. An implicit field assigns a value to each point in 3D space, so that a shape can be extracted as an iso-surface. Our implicit field decoder is trained to perform this assignment by means of a binary classifier. Specifically, it takes a point coordinate, along with a feature vector encoding a shape, and outputs a value which indicates whether the point is outside the shape or not. By replacing conventional decoders by our decoder for representation learning and generative modeling of shapes, we demonstrate superior results for tasks such as shape autoencoding, generation, interpolation, and single-view 3D reconstruction, particularly in terms of visual quality.
Deep learning has yielded state-of-the-art performance on many natural language processing tasks including named entity recognition (NER). However, this typically requires large amounts of labeled data. In this work, we demonstrate that the amount of labeled training data can be drastically reduced when deep learning is combined with active learning. While active learning is sample-efficient, it can be computationally expensive since it requires iterative retraining. To speed this up, we introduce a lightweight architecture for NER, viz., the CNN-CNN-LSTM model consisting of convolutional character and word encoders and a long short term memory (LSTM) tag decoder. The model achieves nearly state-of-the-art performance on standard datasets for the task while being computationally much more efficient than best performing models. We carry out incremental active learning, during the training process, and are able to nearly match state-of-the-art performance with just 25\% of the original training data.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191