IBM (United States)

CURE2026 workshop provides a venue for research on how interactive and intelligent interfaces can be leveraged to ensure effective uncertainty communication. With the increased capabilities of black-box AI models, particularly large language models (LLMs), there is a growing urgency to understand exactly what these models can and cannot do. Effective uncertainty communication can help users calibrate their trust and reliance on the AI system. However, effectively conveying uncertainty, be it through conversation or by visualizing their confidence levels, remains a major challenge, as users often get confused or misinterpret uncertainty information due to poorly designed interfaces. CURE2026 workshop brings together researchers from HCI, AI, design, and psychology to investigate how we can design, develop, and evaluate intelligent and interactive interfaces for uncertainty communication. The workshop includes a mini-conference section aimed at inspiring discussion and investigating the role of intelligent and interactive interfaces in uncertainty communication. The workshop also features keynotes by Professors Jessica Hullman and Katrien Verbert. Finally, a part of the workshop is dedicated to an interactive activity to discuss and develop guidelines for communicating uncertainty across different domains and tasks.

This study examines the critical role of work-life balance (WLB) within the high-pressure environment of the logistics industry. Characterized by demanding schedules, shift work, and rigorous delivery deadlines, the logistics sector presents unique challenges that often blur the lines between professional obligations and personal life. Work-life balance is defined as an individual’s ability to harmonize professional duties with family and social commitments. The research highlights that while a healthy balance fosters increased job satisfaction, motivation, and organizational loyalty, an imbalance often triggers chronic stress, burnout, and diminished productivity. This paper explores current WLB practices, identifies the primary factors influencing equilibrium in logistics firms, and evaluates the resulting levels of employee satisfaction.

National privacy laws diverge between the European Union and United States, hindering transatlantic health data exchange and slowing AI-driven medical innovation. In response, the German Ministry of Health launched the pre-competitive Data for Health initiative, leading to the BRIDGE Pilot Study (2023-2025), a researcher-led effort to address this regulatory and legal gap. Using a mixed-methods approach, including structured surveys (n = 56 expert responses), ranking of steps via relative importance indexing, and 4 Delphi meetings, experts co-developed a practical framework composed of 30 steps in 3 consecutive phases for legally compliant and technically interoperable EU-US health data collaboration. The framework emphasizes early data protection assessments, secure transfer protocols, and iterative governance checks. The final consensus framework provides a stepwise guide to navigate regulatory and legal complexities and operationalize cross-border research. Ongoing input from researchers and stakeholders will help ensure the framework remains adaptable and provides a clear, scalable foundation for cross-border health data exchange.

Modern AI accelerators adopt dataflow architectures to achieve both high peak throughput (TOPS) and energy efficiency (TOPS/W). These designs feature wide datapaths and hierarchical scratchpad memories that supply dense compute arrays with high-bandwidth data access and extensive operand reuse. Complementing the compute–memory subsystem is a lightweight control path that orchestrates data movement, program loading, and register initialization. To reduce energy and area overheads, conventional processor features—such as instruction caches, execution stacks, and branch speculation—are deliberately omitted. While this streamlined design maximizes efficiency, it shifts a critical responsibility onto the compiler: transforming complex kernels into highly compact instruction streams that must fit entirely within the limited instruction buffers (IBUFFs) of the accelerator’s programmable units.In this paper, we introduce two novel compiler transformations—Loop Absorption (LA) and Loop Index Set Merging (LISM) for ultra compact code generation. Loop Absorption merges isomorphic sibling operations into a single loop body, while LISM unifies adjacent loops with similar bodies into a unified iteration space. Together, these complementary techniques eliminate redundant code patterns and produce compact hierarchical loop nests. We implement LA and LISM in the IBM Spyre compiler and evaluate them on diverse deep learning workloads including ResNet-50, Inception-v3, SSD, and BERT-Large. Across these models, our combined approach achieves a geometric mean compression of 1.48× over the baseline, enabling layers that previously exceeded IBUFF capacity to compile successfully.

Image-based neural networks are widely used across diverse applications, yet their inherent susceptibility to backdoor attacks has raised growing concerns. In particular, data-free backdoor attacks, which exploit benign but redundant pathways to manipulate model predictions, pose significant risks. Existing defense methods rely heavily on conventional pruning strategies based on weight magnitude or neuron activation, which often cause excessive distortion of the original representations and lead to substantial loss of useful information. In this work, we propose a pruning-based defense method ALobot that leverages a label-dependence pruning. This method is especially well-suited for mitigating backdoors in models deployed to edge devices, where computational resources are constrained and defenses must lower attack success without imposing heavy overhead or degrading model performance. Specifically, we selectively prune neurons that contribute substantially to inter-class discrimination but minimally to intra-class consistency, achieving a noticeable reduction in backdoor attack success rate without sacrificing model performance through a lightweight mitigation. The initial results show that ALobot achieves a reduction in attack success rate by 2% to 8%, while maintaining a high clean performance and requiring modification of only 1/20 of the parameters compared to conventional L1-norm pruning and L2-norm pruning methods.

The accumulation of municipal solid waste (MSW) continues to rise due to burgeoning population, rapid global urbanization and economic growth, intensifying ecological concerns associated with landfills and greenhouse gas (GHG) emissions. Over the past 2 decades, global waste generation has surged by 50%, with one-third remaining uncollected and about 70% sent to landfills. This review examines the critical role of integrating emerging technologies, such as advanced sensors and artificial intelligence (AI), into end-to-end MSW management to alleviate landfill burdens. The suitability of various AI tools for different stages of MSW management is assessed, alongside the deployment of advanced sensors including hyperspectral cameras, computer vision systems, and internet of things (IoT) devices for material identification. Applications of genetic algorithms and reinforcement learning for optimizing collection routes, reducing costs, and lowering emissions are highlighted. Life cycle assessment (LCA) across all stages of MSW management is also reviewed, along with future trends in leveraging generative AI, natural language processing (NLP), and agent-based AI systems to analyze waste generation patterns and public sentiment. Efficient collection and handling can be enhanced through route optimization with geographic information systems and real-time bin-level monitoring. Furthermore, sensor-embedded, real-time object detection systems paired with robotics enable material characterization and automated sorting, thereby lowering costs and diverting waste from landfills into value-added products for diverse industrial sectors including packaging, chemicals, textiles, metals and glass, transportation, and electronics industries. Without intervention, global waste is projected to reach 4.54 billion tons by 2050, contributing direct economic costs of $400 billion and roughly 2.38 billion tons of CO 2 -equivalent emissions annually. This review demonstrates how AI-driven, end-to-end solutions for MSW management can mitigate economic and environmental challenges, while directly supporting the United Nations Sustainable Development (UNDP) goals related to innovation and infrastructure (SDG 9), sustainable cities (SDG 11), responsible consumption and production (SDG 12), and climate action (SDG 13).

As the quality of code generated by Large Language Models (LLMs) improves, their adoption in the software industry for automated code generation continues to grow. Researchers primarily focus on enhancing the functional correctness of the generated code while commonly overlooking its energy efficiency and environmental impact. This paper investigates the energy efficiency of the code generated by 20 popular LLMs for 878 programming problems of varying difficulty levels and diverse algorithmic categories selected from the LeetCode platform by comparing them against canonical human-written solutions. Although LLMs can produce functionally correct results in most cases, our findings show that the performance and energy efficiency of LLM-produced solutions are often far below those of human-written solutions. Among the studied LLMs, DeepSeek-v3 and GPT-4o generate the most energy-efficient code, whereas Grok-2 and Gemini-1.5-Pro are among the least energy-efficient models. On average, human-generated canonical solutions are approximately 1.17 times more energy efficient than DeepSeek-v3, 1.21 times more energy efficient than GPT-4o, and over 2 times more energy efficient than Grok-2 and Gemini-1.5-Pro. For specific algorithmic groups such as dynamic programming, backtracking, and bit manipulation, LLM-generated code can consume up to 450 times more energy than human-generated canonical solutions.

Introduction Plasma Enhanced Chemical Vapor Deposition (PECVD) has become the predominant dielectric deposition process chosen by the nanoelectronics fabrication industry due to high throughput, low temperature processing and process controllability for a majority of applications and chemistries. This is especially true for the last 25 years with Cu-Low k dielectric including inter-level dielectric, diffusion barriers and hard masks. PECVD does however have innate difficulties in thickness control, uniformity and reproducibility in depositing films in the ultrathin deposition regime (3-10 nm), especially in large 300 mm plasma CVD reactors. For thin film deposition, it is in this ultrathin regime that the initial transient (ITS) phenomena (1,2), nucleation phase, and Plasma stabilization phases of the deposition process most heavily impact the resulting process control and film properties, with the highest impact, and greatest challenges seen at step deposition times (SDT) below 1 second. The ever-shrinking dimensions of semiconductor devices is pushing the boundary of what the historical PECVD processes are capable of, necessitating new developments to enable and prolong these processes into the future. In this paper, the development of “Pulsed Plasma CVD” will be presented to extend the lifetime of the current generation of PECVD equipment, and its relevance into the sub nm technology nodes. Experimental and Discussion The ultrathin dielectrics were deposited sequentially in commercial 300 mm PECVD four-station systems using various chemistries SiH4/N2O /NH3 or TMS/ NH3 at 350-400C, 13.56 MHz RF frequencies. UV cure was used on the layer deposited at the lower temperature. Films were characterized by using various MIM/MIS electrical measurement and other analysis techniques including nano-indentation for modulus (E) and hardness (H) to determine the film electrical and mechanical properties. Other characterization methods, ellipsometry, Fourier Transformed Infrared (FTIR) spectroscopy, X-Ray electron spectroscopy (XPS) was used for the characterization of the bonding structure. and depth profiles measure film’s chemical composition. The key developments that will be discussed center around deposition breaks and sequential partial depositions taking place individually in all four sub-stations. By taking advantage of the ITS phenomena multiple times throughout the deposition, pseudo-independent layers are deposited sequentially forming a stack, wherein the individual layers can be detected through advanced TEM analysis (3) and high sensitivity, low sputter rate SIMS. With the enhanced automatic matching capabilities of current generation equipment, what we have found is that the step deposition time required for plasma stability can be significantly shortened and transitioned into an PE-ALD like state. By doing so the deposition growth rate can be controlled by the number of sequential partial depositions taking place and the significantly shortened SDT. These findings were then transferred to a Nitrogen doped Silicon Carbide PECVD process on a similar platform, with the goal of improving the oxidation barrier properties of the film and further reducing the required film thickness required to maintain adequate diffusion barrier properties. By minimizing the SDT, and significantly increasing the number of cycles, and by increasing the number of times that it is employed the ITS phenomena can be optimized resulting in a thinner, more effective oxidation barrier than previously thought (2) without sacrificing thickness, uniformity control or process repeatability (Figures 1,2,3). By employing a similar flow as the damascene Cu process used in advanced BEOL fabrication, the ultra-thin multilayer SiCN films are deposited on blanket 300mm Silicon wafers with copper and a thin cobalt cap. To evaluate the capping capability and reliability, the wafers undergo multiple rounds of annealing at increased temperatures while in an oxygen environment. Observations for copper oxidation are made between each round. The result is that by increasing the number of cyclic depositions at vastly reduced cycle deposition times, the “Pulsed Plasma CVD” SiCN cap can successfully prevent copper oxidation at film thicknesses 40% less than the base process while maintaining or improving as-deposited uniformity control (Figures 4,5). These advances are made possible by the advanced automatic RF matching capabilities implemented by the equipment vendors, and the improvement of these features will be necessary to continually push the limits that were inherent to previous generations of PECVD Equipment. Summary In summary, by transitioning standard PECVD processes to a Pulsed Plasma CVD scheme, improved thickness control can be attained in the ultrathin regime by the use of advanced RF control systems. These cyclic processes can then be further thinned down while improving the barrier properties to levels previously unattainable using standard non-cyclic PECVD by making repeated use of the initial transient phase of the deposition process. Figure 1