Large Language Models (LLMs) can achieve near-optimal lossless compression by acting as powerful probability models. We investigate their use in the lossy domain, where reconstruction fidelity is traded for higher compression ratios. This paper introduces Error-Bounded Predictive Coding (EPC), a lossy text codec that leverages a Masked Language Model (MLM) as a decompressor. Instead of storing a subset of original tokens, EPC allows the model to predict masked content and stores minimal, rank-based corrections only when the model's top prediction is incorrect. This creates a residual channel that offers continuous rate-distortion control. We compare EPC to a simpler Predictive Masking (PM) baseline and a transform-based Vector Quantisation with a Residual Patch (VQ+RE) approach. Through an evaluation that includes precise bit accounting and rate-distortion analysis, we demonstrate that EPC consistently dominates PM, offering superior fidelity at a significantly lower bit rate by more efficiently utilising the model's intrinsic knowledge.

Predictive Coding (PC) offers a biologically plausible alternative to backpropagation for neural network training, yet struggles with deeper architectures. This paper identifies the root cause and provides a principled solution. We uncover that the canonical state-based formulation of PC (sPC) is, by design, deeply inefficient on digital hardware, due to an inherent signal decay problem that scales exponentially with depth. To address this fundamental limitation, we introduce a novel reparameterization of PC, named error-based PC (ePC), which does not suffer from signal decay. By optimizing over prediction errors rather than states, ePC enables signals to reach all layers simultaneously and unattenuated, converging orders of magnitude faster than sPC. Experiments across multiple architectures and datasets demonstrate that ePC matches backpropagation's performance even for deeper models where sPC struggles. Besides practical improvements, our work provides theoretical insight into PC dynamics and establishes a foundation for scaling bio-inspired learning to deeper architectures on digital hardware and beyond.

UK city-dwellers may have spotted a strangely shaped car cruising around their neighbourhood earlier this year. It looks just like a Google Street View vehicle, with a camera rig emerging from the back end to scan its environment – and like the Google cars, it, too, is scanning and photographing city streets. But these modified Teslas aren't just taking photos. They are kitted out with state-of-the-art sensors and scanners that enable them to report back on the exact dimensions, heat loss, materials, age and state of dilapidation of every building they drive past. Armed with this so-called built environment scanning system (BESS), the cars have been on the hunt to find out how leaky and run-down the UK's building stock really is.

The usage of federated learning (FL) in Vehicular Ad hoc Networks (VANET) has garnered significant interest in research due to the advantages of reducing transmission overhead and protecting user privacy by communicating local dataset gradients instead of raw data. However, implementing FL in VANETs faces challenges, including limited communication resources, high vehicle mobility, and the statistical diversity of data distributions. In order to tackle these issues, this paper introduces a novel framework for hierarchical federated learning (HFL) over multi-hop clustering-based VANET. The proposed method utilizes a weighted combination of the average relative speed and cosine similarity of FL model parameters as a clustering metric to consider both data diversity and high vehicle mobility. This metric ensures convergence with minimum changes in cluster heads while tackling the complexities associated with non-independent and identically distributed (non-IID) data scenarios. Additionally, the framework includes a novel mechanism to manage seamless transitions of cluster heads (CHs), followed by transferring the most recent FL model parameter to the designated CH. Furthermore, the proposed approach considers the option of merging CHs, aiming to reduce their count and, consequently, mitigate associated overhead. Through extensive simulations, the proposed hierarchical federated learning over clustered VANET has been demonstrated to improve accuracy and convergence time significantly while maintaining an acceptable level of packet overhead compared to previously proposed clustering algorithms and non-clustered VANET.

In a previous article of February 6, 2020, we discussed the EPO Receiving Section's refusal, in January 2020, of two European patent applications where an AI system called DABUS was indicated as the inventor1 . We then looked at the grounds of the decisions2 (concerning applications EP 18 275 163 and EP 18 275 174 for "food container" and "devices and methods for attracting enhanced attention"), and predicted that the EPO Board of Appeal (BoA) was bound to shed light on the novel and intriguing legal issue of whether a non-human, such as an artificial intelligence (AI), could be named as inventor in the system of the EPC. The BoA has now issued its decision, which is worth commenting. The applicant, one Mr. Stephen Thaler, had filed his appeals against the refusal (cases J 8/20 and J 9/20), along with an auxiliary request whereby no person was allegedly identified as inventor, but a natural person was indicated to hold "the right to the European Patent by virtue of being the owner and creator of" the DABUS AI system. By decision of December 21, 20213, the BoA dismissed the appeal, confirming that the EPC required the inventor to be a person with legal capacity.

Evolved packet core (EPC, Figure 5) is a distributed system of different nodes, each consisting of diverse network functions (NFs) that are required to manage the LTE network. The EPC consists of data and control data planes: the data plane enforces operator policies (e.g., DPI, QoS classes, and accounting) on data traffic to/from the user equipment (UE), whereas the control plane provides key control and management functions such as access control, mobility, and security management.

In multi-agent games, the complexity of the environment can grow exponentially as the number of agents increases, so it is particularly challenging to learn good policies when the agent population is large. In this paper, we introduce Evolutionary Population Curriculum (EPC), a curriculum learning paradigm that scales up Multi-Agent Reinforcement Learning (MARL) by progressively increasing the population of training agents in a stage-wise manner. Furthermore, EPC uses an evolutionary approach to fix an objective misalignment issue throughout the curriculum: agents successfully trained in an early stage with a small population are not necessarily the best candidates for adapting to later stages with scaled populations. Concretely, EPC maintains multiple sets of agents in each stage, performs mix-and-match and fine-tuning over these sets and promotes the sets of agents with the best adaptability to the next stage. We implement EPC on a popular MARL algorithm, MADDPG, and empirically show that our approach consistently outperforms baselines by a large margin as the number of agents grows exponentially. The project page is https://sites.google.com/view/epciclr2020.

Encoder Products Company (EPC) is a leading designer and world-wide manufacturer of motion sensing devices. EPC began operations in 1969, producing a line of custom encoders (the original Cube series) from a small, home-based shop. Today, EPC is the largest privately-held encoder manufacturer in North America, producing the most complete line of incremental and absolute rotary encoders in the industry. Meeting the diverse needs of a wide range of global customers, EPC's core philosophy is that each and every customer deserves quality products, superior customer service, and expert support. Adherence to these principals has enabled EPC to achieve its goal of maintaining long-lasting customer relationships.

Ford CEO Mark Fields holds Velodyne Puck LIDAR sensor at a press conference at CES in Las Vegas in January. Carmakers and tech firms competing to develop automated vehicles seek a combination of sensors and cameras that provide maximum perception and visibility of surroundings at a cost that's manageable for mass production. Velodyne, a leading maker of laser-based LiDAR, or Light, Detection and Ranging, sensors, says it has designed a new solid-state version of its technology that provides 3D imaging for automated vehicle systems that will cost less than $50 per unit when manufactured at high volume. That's a fraction of the $8,000 cost of its current mechanical spinning LIDAR devices used in prototype robotic cars. The new design "creates a true solid-state LiDAR sensor, while significantly raising the bar as to what can be expected from LiDAR sensors as far as cost, size and reliability," company founder and CEO David Hall said in a statement.

In this paper we examine the practical use of hardware neural networks in an autonomous mobile robot. We have developed a hardware neural system based around a custom VLSI chip, EP SILON III, designed specifically for embedded hardware neural applications. We present here a demonstration application of an autonomous mobile robot that highlights the flexibility of this system.