Zoi Kaoudi

Alexander Renz-Wieland, Andreas Kieslinger, Robert Gericke et al.

Parameter management is essential for distributed training of large machine learning (ML) tasks. Some ML tasks are hard to distribute because common approaches to parameter management can be highly inefficient. Advanced parameter management approaches -- such as selective replication or dynamic parameter allocation -- can improve efficiency, but to do so, they typically need to be integrated manually into each task's implementation and they require expensive upfront experimentation to tune correctly. In this work, we explore whether these two problems can be avoided. We first propose a novel intent signaling mechanism that integrates naturally into existing ML stacks and provides the parameter manager with crucial information about parameter accesses. We then describe AdaPM, a fully adaptive, zero-tuning parameter manager based on this mechanism. In contrast to prior systems, this approach separates providing information (simple, done by the task) from exploiting it effectively (hard, done automatically by AdaPM). In our experimental evaluation, AdaPM matched or outperformed state-of-the-art parameter managers out of the box, suggesting that automatic parameter management is possible.

Zoi Kaoudi, Abelardo Carlos Martinez Lorenzo, Volker Markl

Knowledge graph completion (a.k.a.~link prediction), i.e.,~the task of inferring missing information from knowledge graphs, is a widely used task in many applications, such as product recommendation and question answering. The state-of-the-art approaches of knowledge graph embeddings and/or rule mining and reasoning are data-driven and, thus, solely based on the information the input knowledge graph contains. This leads to unsatisfactory prediction results which make such solutions inapplicable to crucial domains such as healthcare. To further enhance the accuracy of knowledge graph completion we propose to loosely-couple the data-driven power of knowledge graph embeddings with domain-specific reasoning stemming from experts or entailment regimes (e.g., OWL2). In this way, we not only enhance the prediction accuracy with domain knowledge that may not be included in the input knowledge graph but also allow users to plugin their own knowledge graph embedding and reasoning method. Our initial results show that we enhance the MRR accuracy of vanilla knowledge graph embeddings by up to 3x and outperform hybrid solutions that combine knowledge graph embeddings with rule mining and reasoning up to 3.5x MRR.

Alexander Renz-Wieland, Rainer Gemulla, Zoi Kaoudi et al.

Parameter servers (PSs) facilitate the implementation of distributed training for large machine learning tasks. In this paper, we argue that existing PSs are inefficient for tasks that exhibit non-uniform parameter access; their performance may even fall behind that of single node baselines. We identify two major sources of such non-uniform access: skew and sampling. Existing PSs are ill-suited for managing skew because they uniformly apply the same parameter management technique to all parameters. They are inefficient for sampling because the PS is oblivious to the associated randomized accesses and cannot exploit locality. To overcome these performance limitations, we introduce NuPS, a novel PS architecture that (i) integrates multiple management techniques and employs a suitable technique for each parameter and (ii) supports sampling directly via suitable sampling primitives and sampling schemes that allow for a controlled quality--efficiency trade-off. In our experimental study, NuPS outperformed existing PSs by up to one order of magnitude and provided up to linear scalability across multiple machine learning tasks.

Aisha Mohamed, Ghadeer Abuoda, Abdurrahman Ghanem et al.

Knowledge graphs represented as RDF datasets are integral to many machine learning applications. RDF is supported by a rich ecosystem of data management systems and tools, most notably RDF database systems that provide a SPARQL query interface. Surprisingly, machine learning tools for knowledge graphs do not use SPARQL, despite the obvious advantages of using a database system. This is due to the mismatch between SPARQL and machine learning tools in terms of data model and programming style. Machine learning tools work on data in tabular format and process it using an imperative programming style, while SPARQL is declarative and has as its basic operation matching graph patterns to RDF triples. We posit that a good interface to knowledge graphs from a machine learning software stack should use an imperative, navigational programming paradigm based on graph traversal rather than the SPARQL query paradigm based on graph patterns. In this paper, we present RDFFrames, a framework that provides such an interface. RDFFrames provides an imperative Python API that gets internally translated to SPARQL, and it is integrated with the PyData machine learning software stack. RDFFrames enables the user to make a sequence of Python calls to define the data to be extracted from a knowledge graph stored in an RDF database system, and it translates these calls into a compact SPQARL query, executes it on the database system, and returns the results in a standard tabular format. Thus, RDFFrames is a useful tool for data preparation that combines the usability of PyData with the flexibility and performance of RDF database systems.

Jonas Traub, Jorge-Arnulfo Quiané-Ruiz, Zoi Kaoudi et al.

Data, algorithms, and compute/storage infrastructure are key assets that drive data science and artificial intelligence applications. As providing all these assets requires a huge investment, data science and artificial intelligence technologies are currently dominated by a small number of providers who can afford these investments. This leads to lock-in effects and hinders features that require a flexible exchange of assets among users. In this vision paper, we present Agora, a unified asset ecosystem. The Agora system provides the technical infrastructure that allows for offering and using data and algorithms, as well as physical infrastructure components. Agora is designed as an open ecosystem of asset marketplaces and provides to a broad audience not only data but the entire data value chain (including computational resources and human expertise). Agora (i) leverages a fine-grained exchange of assets, (ii) allows for combining assets to novel applications, and (iii) flexibly executes such applications on available resources. As a result, Agora overcomes lock-in effects and removes entry barriers for new asset providers. In contrast to existing data management systems, Agora operates in a heavily decentralized and dynamic environment: Data, algorithms, and even compute resources are dynamically created, modified, and removed by different stakeholders. Agora presents novel research directions for the data management community as a whole: It requires to combine our traditional expertise in scalable data processing and management with infrastructure provisioning as well as economic and application aspects of data, algorithms, and infrastructure.

Stefano Giovanni Rizzo, Ji Lucas, Zoi Kaoudi et al.

We propose AI-CARGO, a revenue management system for air-cargo that combines machine learning prediction with decision-making using mathematical optimization methods. AI-CARGO addresses a problem that is unique to the air-cargo business, namely the wide discrepancy between the quantity (weight or volume) that a shipper will book and the actual received amount at departure time by the airline. The discrepancy results in sub-optimal and inefficient behavior by both the shipper and the airline resulting in the overall loss of potential revenue for the airline. AI-CARGO also includes a data cleaning component to deal with the heterogeneous forms in which booking data is transmitted to the airline cargo system. AI-CARGO is deployed in the production environment of a large commercial airline company. We have validated the benefits of AI-CARGO using real and synthetic datasets. Especially, we have carried out simulations using dynamic programming techniques to elicit the impact on offloading costs and revenue generation of our proposed system. Our results suggest that combining prediction within a decision-making framework can help dramatically to reduce offloading costs and optimize revenue generation.