AutoVideo: An Automated Video Action Recognition System

Zha, Daochen; Bhat, Zaid Pervaiz; Chen, Yiwei; Wang, Yicheng; Ding, Sirui; Chen, Jiaben; Lai, Kwei-Herng; Bhat, Mohammad Qazim; Jain, Anmoll Kumar; Reyes, Alfredo Costilla; Zou, Na; Hu, Xia

doi:10.24963/ijcai.2022/862

Cited by 9 publications

(11 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It remains a challenge to systematically and simultaneously tackle data issues across multiple DCAI tasks. AutoML could be one of the promising directions to approach this goal with end-to-end data pipeline search [8,12,7,18,45].…”

Section: Open Research Challengesmentioning

confidence: 99%

Data-centric AI: Perspectives and Challenges

Zha¹,

Bhat²,

Lai³

et al. 2023

Preprint

View full text Add to dashboard Cite

The role of data in building AI systems has recently been significantly magnified by the emerging concept of data-centric AI (DCAI), which advocates a fundamental shift from model advancements to ensuring data quality and reliability. Although our community has continuously invested efforts into enhancing data in different aspects, they are often isolated initiatives on specific tasks. To facilitate the collective initiative in our community and push forward DCAI, we draw a big picture and bring together three general missions: training data development, evaluation data development, and data maintenance. We provide a top-level discussion on representative DCAI tasks and share perspectives. Finally, we list open challenges to motivate future exploration.

show abstract

Section: Open Research Challengesmentioning

confidence: 99%

Data-centric AI: Perspectives and Challenges

Zha¹,

Bhat²,

Lai³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…AutoML. AutoML systems have achieved remarkable success in various ML design tasks, such as hyperparameter tuning [13,33], algorithm selection [35,39], neural architecture search [21,70,71, 54, 29], meta-learning [15,50,3], and pipeline search [49,12,18,11,56,44,20,68,37,24,31,30,32,25].…”

Section: Related Workmentioning

confidence: 99%

“…Feature preprocessing is so important that around 50% of the time is spent on data preprocessing in building an ML system, reported in a survey collected from practitioners [40]. Thus, modern automated machine learning (AutoML) systems have included various preprocessing primitives in building ML pipelines [11,18,27,36,44,24,68].…”

Section: Introductionmentioning

confidence: 99%

Towards Personalized Preprocessing Pipeline Search

Martinez¹,

Zha²,

Tan³

et al. 2023

Preprint

View full text Add to dashboard Cite

Feature preprocessing, which transforms raw input features into numerical representations, is a crucial step in automated machine learning (AutoML) systems. However, the existing systems often have a very small search space for feature preprocessing with the same preprocessing pipeline applied to all the numerical features. This may result in sub-optimal performance since di erent datasets often have various feature characteristics, and features within a dataset may also have their own preprocessing preferences. To bridge this gap, we explore personalized preprocessing pipeline search, where the search algorithm is allowed to adopt a di erent preprocessing pipeline for each feature. This is a challenging task because the search space grows exponentially with more features. To tackle this challenge, we propose ClusterP3S, a novel framework for Personalized Preprocessing Pipeline Search via Clustering. The key idea is to learn feature clusters such that the search space can be signi cantly reduced by using the same preprocessing pipeline for the features within a cluster. To this end, we propose a hierarchical search strategy to jointly learn the clusters and search for the optimal pipelines, where the upper-level search optimizes the feature clustering to enable better pipelines built upon the clusters, and the lower-level search optimizes the pipeline given a speci c cluster assignment. We instantiate this idea with a deep clustering network that is trained with reinforcement learning at the upper-level, and random search at the lower level. Experiments on benchmark classi cation datasets demonstrate the e ectiveness of enabling feature-wise preprocessing pipeline search.

show abstract

“…Deep RL has shown promise in accomplishing goaloriented tasks [23,28,30,33,35]. Recently, deep RL has been applied to various machine learning model design tasks, such as neural architecture search [41], pipeline search [18,20,34], data augmentation/sampling [6,31,32]. Our work also falls into this line of studies but we instead focus on optimizing the model efficiency.…”

Section: Related Workmentioning

confidence: 99%

“…Deep RL has recently made significant progress in games [46,47,48,49,50,51,52,53,54]. Our work is related to using RL to optimize machine learning model designs, such as neural architecture search [55,56,57,58,59], data augmentation [60], data sampling [61,62], pipeline search [63,64,65]. However, these methods often only focus on one task and can not generalize to unseen tasks.…”

Section: Related Workmentioning

confidence: 99%

AutoShard: Automated Embedding Table Sharding for Recommender Systems

Zha

Liu

Bhushanam

et al. 2022

Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

Self Cite

View full text Add to dashboard Cite

Embedding learning is an important technique in deep recommendation models to map categorical features to dense vectors. However, the embedding tables often demand an extremely large number of parameters, which become the storage and efficiency bottlenecks. Distributed training solutions have been adopted to partition the embedding tables into multiple devices. However, the embedding tables can easily lead to imbalances if not carefully partitioned. This is a significant design challenge of distributed systems named embedding table sharding, i.e., how we should partition the embedding tables to balance the costs across devices, which is a non-trivial task because 1) it is hard to efficiently and precisely measure the cost, and 2) the partition problem is known to be NP-hard. In this work, we introduce our novel practice in Meta, namely AutoShard, which uses a neural cost model to directly predict the multi-table costs and leverages deep reinforcement learning to solve the partition problem. Experimental results on an open-sourced large-scale synthetic dataset and Meta's production dataset demonstrate the superiority of AutoShard over the heuristics. Moreover, the learned policy of AutoShard can transfer to sharding tasks with various numbers of tables and different ratios of the unseen tables without any fine-tuning. Furthermore, AutoShard can efficiently shard hundreds of tables in seconds. The effectiveness, transferability, and efficiency of AutoShard make it desirable for production use. Our algorithms have been deployed in Meta production environment. A prototype is available at https://github.com/daochenzha/autoshard CCS CONCEPTS• Computing methodologies → Reinforcement learning; Machine learning approaches.

show abstract

AutoVideo: An Automated Video Action Recognition System

Cited by 9 publications

References 7 publications

Data-centric AI: Perspectives and Challenges

Data-centric AI: Perspectives and Challenges

Towards Personalized Preprocessing Pipeline Search

AutoShard: Automated Embedding Table Sharding for Recommender Systems

Contact Info

Product

Resources

About