IBM Research - United Kingdom

A 4.9-6.4-Gb/s two-level SerDes ASIC I/O core employing a four-tap feed-forward equalizer (FFE) in the transmitter and a five-tap decision-feedback equalizer (DFE) in the receiver has been designed in 0.13-/spl mu/m CMOS. The transmitter features a total jitter (TJ) of 35 ps p-p at 10/sup -12/ bit error rate (BER) and can output up to 1200 mVppd into a 100-/spl Omega/ differential load. Low jitter is achieved through the use of an LC-tank-based VCO/PLL system that achieves a typical random jitter of 0.6 ps over a phase noise integration range from 6 MHz to 3.2 GHz. The receiver features a variable-gain amplifier (VGA) with gain ranging from -6to +10dB in /spl sim/1dB steps, an analog peaking amplifier, and a continuously adapted DFE-based data slicer that uses a hybrid speculative/dynamic feedback architecture optimized for high-speed operation. The receiver system is designed to operate with a signal level ranging from 50 to 1200 mVppd. Error-free operation of the system has been demonstrated on lossy transmission line channels with over 32-dB loss at the Nyquist (1/2 Bd rate) frequency. The Tx/Rx pair with amortized PLL power consumes 290 mW of power from a 1.2-V supply while driving 600 mVppd and uses a die area of 0.79 mm/sup 2/.

We present a framework that couples a high-fidelity compositional reservoir simulator with Bayesian optimization (BO) for injection well scheduling optimization in geological carbon sequestration. This work represents one of the first at tempts to apply BO and high-fidelity physics models to geological carbon storage. The implicit parallel accurate reservoir simulator (IPARS) is utilized to accurately capture the underlying physical processes during CO2 sequestration. IPARS provides a framework for several flow and mechanics models and thus supports both stand-alone and coupled simulations. In this work, we use the compositional flow module to simulate the geological carbon storage process. The compositional flow model, which includes a hysteretic three-phase relative permeability model, accounts for three major CO2 trapping mechanisms: structural trapping, residual gas trapping, and solubility trapping. Furthermore, IPARS is coupled to the International Business Machines (IBM) Corporation Bayesian Optimization Accelerator (BOA) for parallel optimizations of CO2 injection strategies during field-scale CO2 sequestration. BO builds a probabilistic surrogate for the objective function using a Bayesian machine learning algorithm—the Gaussian process regression, and then uses an acquisition function that leverages the uncertainty in the surrogate to decide where to sample. The IBM BOA addresses the three weaknesses of standard BO that limits its scalability in that IBM BOA supports parallel (batch) executions, scales better for high-dimensional problems, and is more robust to initializations. We demonstrate these merits by applying the algorithm in the optimization of the CO2 injection schedule in the Cranfield site in Mississippi, USA, using field data. The optimized injection schedule achieves 16% more gas storage volume and 56% less water/surfactant usage compared with the baseline. The performance of BO is compared with that of a genetic algorithm (GA) and a covariance matrix adaptation (CMA)-evolution strategy (ES). The results demonstrate the superior performance of BO, in that it achieves a competitive objective function value with over 60% fewer forward model evaluations.

Abstract In elite sports, there is an opportunity to take advantage of rich and detailed datasets generated across multiple threads of the sporting business. Challenges currently exist due to time constraints to analyse the data, as well as the quantity and variety of data available to assess. Artificial Intelligence (AI) techniques can be a valuable asset in assisting decision makers in tackling such challenges, but deep AI skills are generally not held by those with rich experience in sporting domains. Here, we describe how certain commonly available AI services can be used to provide analytic assistance to sports experts in exploring, and gaining insights from, typical data sources. In particular, we focus on the use of Natural Language Processing and Conversational Interfaces to provide users with an intuitive and time‐saving toolkit to explore their datasets and the conclusions arising from analytics performed on them. We show the benefit of presenting powerful AI and analytic techniques to domain experts, showing the potential for impact not only at the elite level of sports, where AI and analytic capabilities may be more available, but also at a more grass‐roots level where there is generally little access to specialist resources. The work described in this paper was trialled with Leatherhead Football Club, a semi‐professional team that, at the time, were based in the English 7th tier of football.

While a significant part of communication in the workplace is now happening online, current platforms don’t fully support socio-cognitive nonverbal communication, which hampers the shared understanding and creativity of virtual teams. Given text-based communication being the main channel for virtual collaboration, we propose a novel solution leveraging an AI-based, dynamic affective recognition system. The app provides live feedback about the affective content of the communication in Slack, in the form of a visual representation and percentage breakdown of the ‘sentiment’ (tone, emoji) and main ‘emotion states’ (e.g. joy, anger). We tested the usability of the app in a quasi-experiment with 30 participants from diverse backgrounds, linguistic analysis and user interviews. The findings show that the app significantly increases shared understanding and creativity within virtual teams. Emerged themes included impression formation assisted by affective recognition, supporting long-term relationships development; identified challenges related to transparency and emotional complexity detected by AI.

With the proliferation of technology, connected and interconnected devices (henceforth referred to as IoT) are fast becoming a viable option to automate the day-to-day interactions of users with their environment—be it manufacturing or home-care automation. However, with the explosion of IoT deployments we have observed in recent years, manually governing the interactions between humans-to-devices—and especially devices-to- devices—is an impractical task, if not an impossible task. This is because devices have their own obligations and prohibitions in context, and humans are not equip to maintain a bird’s-eye-view of the interaction space. Motivated by this observation, in this paper, we propose an end-to-end framework that (a) automatically dis- covers devices, and their associated services and capabilities w.r.t. an ontology; (b) supports representation of high-level—and expressive—user policies to govern the devices and services in the environment; (c) pro- vides efficient procedur es to refine and reason about policies to automate the management of interactions; and (d) delegates similar capable devices to fulfill the interactions, when conflicts occur. We then present our initial work in instrumenting the framework and discuss its details.

Summary Scientific publications from a group or consortium often form a coherent larger body of work with underlying threads and relationships. Rich social, structural, and topical networks between authors and organizations can be identified, and to convey these we have created the publicly available “Science Library” as a user‐centric, interactive portal. A key consideration in this endeavor is rapid and efficient curation of the corpus of publications, both in terms of assuring quality, as well minimizing the effort required. For this to be sustainable it must offer substantial benefits to the community and avoid excessive operational cost through cumbersome or complex processes. We describe the agility of the Science Library implementation as a controlled natural language (CNL) semantic knowledge graph and describe the different roles within the community to ensure efficient curation, validation, and provenance of the content. By describing the process of curation and validation, alongside the CNL‐based definition of the model we show how relatively non‐technical users are able to interact with, and contribute to the Science Library. This provides an extensible approach, initially based around digital library and virtual community capabilities, that can be applied more broadly to support other desired capabilities of Science Gateways.

The way we travel is changing rapidly and Cooperative Intelligent Transportation Systems (C-ITSs) are at the forefront of this evolution. However, the adoption of C-ITSs introduces new risks and challenges, making cybersecurity a top priority for ensuring safety and reliability. Building on this premise, this paper introduces an envisaged Cybersecurity Centre of Excellence (CSCE) designed to bolster researching, testing, and evaluating the cybersecurity of C-ITSs. We explore the design, functionality, and challenges of CSCE's testing facilities, outlining the technological, security, and societal requirements. Through a thorough survey and analysis, we assess the effectiveness of these systems in detecting and mitigating potential threats, highlighting their flexibility to adapt to future C-ITSs. Finally, we identify current unresolved challenges in various C-ITS domains, with the aim of motivating further research into the cybersecurity of C-ITSs.

Graph neural networks have recently met huge success in various inference tasks including materials property prediction amongst many others. Nevertheless, having an inherently locally-based representation capacity as they do, global representation of materials' structures can only only be achieved by expanding the model complexity which in turn scales up training times and memory consumption. In this work we focus on efficiently capturing global interactions ``in-model'', through long-range edge attentions with minimal memory footprints. We introduce a novel ``contextual'' message passing scheme that better captures global interactions by attending on edges from both the local and global environment of each node in an edge-update fashion. The performance of the proposed model (LiCOMPGNN) is tested on a diverse set of materials property prediction benchmarks and demonstrates competitiveness against state-of-the-art models in several prediction tasks whilst being an order of magnitude smaller in terms of trainable parameters. We further augment the framework to a multiplex graph setting for solid-state data with reciprocal space features taken into account in a multimodal message passing regime. We demonstrate the representation capacity of the proposed variant along with others in maintaining supercell invariance in crystalline property prediction tasks.

Current reinforcement learning automated curricu-lum approaches continual learning by updating the environment. The update is often treated as an optimisation problem - with the teacher agent updating the environment to optimise the student's learning. This work proposes an alternative framing of the problem using a game-theoretic formulation. The learning is defined by a leader - follower cooperative game. This formulation provides an approach for multi-agent curriculum learning that improves agent learning and provides more game equilibrium insights. We observed that under this framework, the agents converge faster to perform on the desired outcomes, compared to the reinforcement learning agent baseline.

Snapshot of the code used in the paper:Force-free molecular dynamics through autoregressive equivariant networks Original repository:https://github.com/IBM/trajcast Commit:c52f107a02176671b21a420b2d4c67e0327b04f7 Archived for reproducibility.

The story of machine learning in general, and its application to molecular design in particular, has been a tale of evolving representations of data. Understanding the implications of the use of a particular representation -- including the existence of so-called `activity cliffs' for cheminformatics models -- is the key to their successful use for molecular discovery. In this work we present a physics-inspired methodology which exploits analogies between model response surfaces and energy landscapes to richly describe the relationship between the representation and the model. From these similarities, a metric emerges which is analogous to the commonly used frustration metric from the chemical physics community. This new property shows state-of-the-art prediction of model error, whilst belonging to a novel class of roughness measure that extends beyond the known data allowing the trivial identification of activity cliffs even in the absence of related training or evaluation data.

The hyperbolic diffusion (HD) is a method for changing an equation set, such as the Navier–Stokes equations, to be hyperbolic in form. This can have the profound impact of making the minimum time step scale with 1/h — rather than 1/h^2 — but at the cost of increasing the number of field variables. Additionally, the approach can offer some unique opportunities for convergence acceleration and optimisation. In this talk, I will give a general introduction to the topic and discuss the state of the ongoing research in this area. After making a case for the HD method, I will go on to explore the opportunities for kernel fusion. The hyperbolised 3D artificial compressibility method for Navier-Stokes leads to a system of 13 equations; this poses some unique challenges in GPU computing. To overcome these challenges, some creative approaches have been used in conjunction with GiMMiK to automatically produce more optimal fused kernels.

Carbon capture and storage is part of the roadmap towards net zero for many countries around the world, since emissions from existing infrastructure are close to estimated carbon budgets. To address this problem, currently 87 carbon capture projects are proposed worldwide in the next 10 years. A major class of commercial carbon capture technology involves capture systems using solvents. Commonly carbon capture solvents feature blends of amines and water. Whilst these blends have proved valuable there is an increasing need to identify new candidate molecules which are more efficient and improve performance. Systematic approaches to improve on the current technology are now needed with increasing urgency to expedite the introduction of cutting edge carbon capture methods. Here, we present a chemical space analysis of amines and carbon capture usage. We proceed to show a framework for computational screening relevant to carbon capture solvents. We demonstrate the use of cloud computing, novel molecular representations and machine learning to screen potential candidates. We show the utility of machine learning in this field for high throughput virtual screening with an exemplar application to absorption capacity classification. Additionally, we highlight the need for improved data awareness and accessibility to enable this field to advance at a pace commensurate to its global importance. Our research brings together multiple methods and domains of expertise to accelerate the discovery of carbon capture solvents.

Carbon capture and storage is part of the roadmap towards net zero for many countries around the world, since emissions from existing infrastructure are close to estimated carbon budgets. To address this problem, currently 87 carbon capture projects are proposed worldwide in the next 10 years. A major class of commercial carbon capture technology involves capture systems using solvents. Commonly carbon capture solvents feature blends of amines and water. Whilst these blends have proved valuable there is an increasing need to identify new candidate molecules which are more efficient and improve performance. Systematic approaches to improve on the current technology are now needed with increasing urgency to expedite the introduction of cutting edge carbon capture methods. Here, we present a chemical space analysis of amines and carbon capture usage. We proceed to show a framework for computational screening relevant to carbon capture solvents. We demonstrate the use of cloud computing, novel molecular representations and machine learning to screen potential candidates. We show the utility of machine learning in this field for high throughput virtual screening with an exemplar application to absorption capacity classification. Additionally, we highlight the need for improved data awareness and accessibility to enable this field to advance at a pace commensurate to its global importance. Our research brings together multiple methods and domains of expertise to accelerate the discovery of carbon capture solvents.

In this presentation I will present recent developments in defining flux reconstruction methods on polygons. In particular, I will show how, using summation-by-parts, how an extended range of flux reconstruction schemes can be defined on triangles, quadrilaterals, and for the first time on hexagons. This extended set can then be used to investigate stable spectral differnce schemes. Further research will be presented demonstrating how these methods can be used to define stable schemes on quadrilaterals with alternative polynomial basis. We will introduce the Euclidean order basis and show that similar numerical perofrmance can be acheived with it as with a maximal order basis, but with fewer points per element.

Snapshot of the code used in the paper:Force-free molecular dynamics through autoregressive equivariant networks Original repository:https://github.com/IBM/trajcast Commit:c52f107a02176671b21a420b2d4c67e0327b04f7 Archived for reproducibility.