More Like this

1. Understanding BERT rankers under distillation

   https://doi.org/10.1145/3409256.3409838  

   Gao, Luyu; Dai, Zhuyun; Callan, Jamie (September 2020, ACM SIGIR International Conference on the Theory of Information Retrieval)

   null (Ed.)

   Deep language models, such as BERT pre-trained on large corpora,have given a huge performance boost to state-of-the-art information retrieval ranking systems. Knowledge embedded in such models allows them to pick up complex matching signals between passages and queries. However, the high computation cost during inference limits their deployment in real-world search scenarios. In this paper, we study if and how the knowledge for search within BERT can be transferred to a smaller ranker through distillation.Our experiments demonstrate that it is crucial to use a proper distillation procedure, which produces up to nine times speed upwhile preserving the state-of-the-art performance. 

   more » « less
2. VarCLR: variable semantic representation pre-training via contrastive learning

   https://doi.org/10.1145/3510003.3510162  

   Chen, Qibin; Lacomis, Jeremy; Schwartz, Edward J.; Neubig, Graham; Vasilescu, Bogdan; Le Goues, Claire (May 2022, International Conference on Software Engineering)

   Variable names are critical for conveying intended program behavior. Machine learning-based program analysis methods use variable name representations for a wide range of tasks, such as suggesting new variable names and bug detection. Ideally, such methods could capture semantic relationships between names beyond syntactic similarity, e.g., the fact that the names average and mean are similar. Unfortunately, previous work has found that even the best of previous representation approaches primarily capture "relatedness" (whether two variables are linked at all), rather than "similarity" (whether they actually have the same meaning). We propose VarCLR, a new approach for learning semantic representations of variable names that effectively captures variable similarity in this stricter sense. We observe that this problem is an excellent fit for contrastive learning, which aims to minimize the distance between explicitly similar inputs, while maximizing the distance between dissimilar inputs. This requires labeled training data, and thus we construct a novel, weakly-supervised variable renaming dataset mined from GitHub edits. We show that VarCLR enables the effective application of sophisticated, general-purpose language models like BERT, to variable name representation and thus also to related downstream tasks like variable name similarity search or spelling correction. VarCLR produces models that significantly outperform the state-of-the-art on IdBench, an existing benchmark that explicitly captures variable similarity (as distinct from relatedness). Finally, we contribute a release of all data, code, and pre-trained models, aiming to provide a drop-in replacement for variable representations used in either existing or future program analyses that rely on variable names. 

   more » « less
3. SMART: Robust and Efficient Fine-Tuning for Pre-trained Natural Language Models through Principled Regularized Optimization

   https://doi.org/10.18653/v1/2020.acl-main.197  

   Jiang, Haoming; He, Pengcheng; Chen, Weizhu; Liu, Xiaodong; Gao, Jianfeng; Zhao, Tuo. (July 2020, Annual Meeting of the Association for Computational Linguistics)

   Transfer learning has fundamentally changed the landscape of natural language processing (NLP). Many state-of-the-art models are first pre-trained on a large text corpus and then fine-tuned on downstream tasks. However, due to limited data resources from downstream tasks and the extremely high complexity of pre-trained models, aggressive fine-tuning of- ten causes the fine-tuned model to overfit the training data of downstream tasks and fail to generalize to unseen data. To address such an issue in a principled manner, we propose a new learning framework for robust and efficient fine-tuning for pre-trained models to attain better generalization performance. The pro- posed framework contains two important in- gredients: 1. Smoothness-inducing regulariza- tion, which effectively manages the complex- ity of the model; 2. Bregman proximal point optimization, which is an instance of trust- region methods and can prevent aggressive up- dating. Our experiments show that the pro- posed framework achieves new state-of-the-art performance on a number of NLP tasks includ- ing GLUE, SNLI, SciTail and ANLI. More- over, it also outperforms the state-of-the-art T5 model, which is the largest pre-trained model containing 11 billion parameters, on GLUE. 

   more » « less
4. Activity Recognition in Older Adults with Training Data from Younger Adults: Preliminary Results on in Vivo Smartwatch Sensor Data

   https://doi.org/10.1145/3441852.3476475  

   Fatima, Sabahat (October 2021, ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '21))

   Self-tracking using commodity wearables such as smartwatches can help older adults reduce sedentary behaviors and engage in physical activity. However, activity recognition applications that are typically deployed in these wearables tend to be trained on datasets that best represent younger adults. We explore how our activity recognition model, a hybrid of long short-term memory and convolutional layers, pre-trained on smartwatch data from younger adults, performs on older adult data. We report results on week-long data from two older adults collected in a preliminary study in the wild with ground-truth annotations based on activPAL, a thigh-worn sensor. We find that activity recognition for older adults remains challenging even when comparing our model’s performance to state of the art deployed models such as the Google Activity Recognition API. More so, we show that models trained on younger adults tend to perform worse on older adults. 

   more » « less
5. Adaptive Denoising via GainTuning

   Mohan, Sreyas; Vincent, Joshua L.; Manzorro, Ramon; Crozier, Peter; Fernandez-Granda, Carlos; Simoncelli, Eero (January 2021, Advances in neural information processing systems)

   Deep convolutional neural networks (CNNs) for image denoising are typically trained on large datasets. These models achieve the current state of the art, but they do not generalize well to data that deviate from the training distribution. Recent work has shown that it is possible to train denoisers on a single noisy image. These models adapt to the features of the test image, but their performance is limited by the small amount of information used to train them. Here we propose "GainTuning'', a methodology by which CNN models pre-trained on large datasets can be adaptively and selectively adjusted for individual test images. To avoid overfitting, GainTuning optimizes a single multiplicative scaling parameter (the “Gain”) of each channel in the convolutional layers of the CNN. We show that GainTuning improves state-of-the-art CNNs on standard image-denoising benchmarks, boosting their denoising performance on nearly every image in a held-out test set. These adaptive improvements are even more substantial for test images differing systematically from the training data, either in noise level or image type. We illustrate the potential of adaptive GainTuning in a scientific application to transmission-electron-microscope images, using a CNN that is pre-trained on synthetic data. In contrast to the existing methodology, GainTuning is able to faithfully reconstruct the structure of catalytic nanoparticles from these data at extremely low signal-to-noise ratios. 

   more » « less