ActBlue remains under house investigation for fraud allegations as Alexandria Ocasio-Cortez, Bernie Sanders and Abdul El-Sayed continue fundraising through the platform.

Federated Learning (FL) enables the training of machine learning models across decentralized clients while preserving data privacy. However, the presence of anomalous or corrupted clients - such as those with faulty sensors or non representative data distributions - can significantly degrade model performance. Detecting such clients without accessing raw data remains a key challenge. We propose WAFFLE (Wavelet and Fourier representations for Federated Learning) a detection algorithm that labels malicious clients {\it before training}, using locally computed compressed representations derived from either the Wavelet Scattering Transform (WST) or the Fourier Transform. Both approaches provide low-dimensional, task-agnostic embeddings suitable for unsupervised client separation. A lightweight detector, trained on a distillated public dataset, performs the labeling with minimal communication and computational overhead. While both transforms enable effective detection, WST offers theoretical advantages, such as non-invertibility and stability to local deformations, that make it particularly well-suited to federated scenarios. Experiments on benchmark datasets show that our method improves detection accuracy and downstream classification performance compared to existing FL anomaly detection algorithms, validating its effectiveness as a pre-training alternative to online detection strategies.

Millions of people around the world need assistance with feeding. Robotic feeding systems offer the potential to enhance autonomy and quality of life for individuals with impairments and reduce caregiver workload. However, their widespread adoption has been limited by technical challenges such as estimating bite timing, the appropriate moment for the robot to transfer food to a user's mouth. In this work, we introduce WAFFLE: Wearable Approach For Feeding with LEarned bite timing, a system that accurately predicts bite timing by leveraging wearable sensor data to be highly reactive to natural user cues such as head movements, chewing, and talking. We train a supervised regression model on bite timing data from 14 participants and incorporate a user-adjustable assertiveness threshold to convert predictions into proceed or stop commands. In a study with 15 participants without motor impairments with the Obi feeding robot, WAFFLE performs statistically on par with or better than baseline methods across measures of feeling of control, robot understanding, and workload, and is preferred by the majority of participants for both individual and social dining. We further demonstrate WAFFLE's generalizability in a study with 2 participants with motor impairments in their home environments using a Kinova 7DOF robot. Our findings support WAFFLE's effectiveness in enabling natural, reactive bite timing that generalizes across users, robot hardware, robot positioning, feeding trajectories, foods, and both individual and social dining contexts.

It was just before lunchtime on its third day of operation, and the line outside Elon Musk's new Tesla Diner in Hollywood already stretched to nearly 100 people. The restaurant has been billed as a "retro-futuristic" drive-in where you can grab a high-end burger and watch classic films on giant screens, all while charging your Tesla. After months of buildup and controversy, the diner had suddenly opened on Monday, at 4.20pm, the kind of stoner boy joke that Musk is well-known for. Hundreds of fans lined up to try burgers in Cybertruck-shaped boxes, or take photos of the Optimus robot serving popcorn on the roof deck of the gleaming circular diner. But that was for the grand opening.

I just told everyone to keep their bodies to themselves in the car, and then you put your feet on the back of your brother's head, and we see you're on your phone, which we repeatedly asked you to leave at home, and so now there's going to have to be a big consequence, and now a chasm has opened between my consciousness and the words emerging from my mouth, and I hear a cascade of scolding clichés rush forth in a frictionless flow, as if I'm an A.I. chatbot with the prompt "Lecture my kids in a style that they will completely ignore and will cause me deep sadness," because I don't know where all this boilerplate hectoring comes from, but the reason we keep our bodies to ourselves is that we treat our bodies and other people's bodies with respect, and if you keep doing that we're going to tell Nana how you behaved. Could Nana be the one who planted this forest of platitudes in my brain, where it silently germinated until the moment when--stop that right now, we told you that word is inappropriate, and it's even more inappropriate to sing it repeatedly as a catchy jingle so that your brother remembers it and repeats it in the Fives Room at preschool, so if we hear it again it means we have a listening problem, and it means that at some point I must have unwittingly memorized a book titled "Empty Threats for Desperate Weenies." All I know is that if we don't start improving our rule following we're going to start examining why we say everything in the first-person plural, because we sure seem afraid of the implications of saying that it is you who have upset me and that I have decided to enforce a boundary that might cause you unhappiness, and that's why you're going to lose Switch for a week, or at least I'll hide MLB: The Show under an old nasal-strip box in the nightstand and then forget where I put it. I'll get you an ice cream if you don't complain about swim lessons for the rest of the year. Definitely no energy drink called Bust, created by a TikTok M.M.A. fighter with a blue Lambo.

In domains where data are sensitive or private, there is great value in methods that can learn in a distributed manner without the data ever leaving the local devices. In light of this need, federated learning has emerged as a popular training paradigm. However, many federated learning approaches trade transmitting data for communicating updated weight parameters for each local device. Therefore, a successful breach that would have otherwise directly compromised the data instead grants whitebox access to the local model, which opens the door to a number of attacks, including exposing the very data federated learning seeks to protect. Additionally, in distributed scenarios, individual client devices commonly exhibit high statistical heterogeneity. Many common federated approaches learn a single global model; while this may do well on average, performance degrades when the i.i.d. assumption is violated, underfitting individuals further from the mean, and raising questions of fairness. To address these issues, we propose Weight Anonymized Factorization for Federated Learning (WAFFLe), an approach that combines the Indian Buffet Process with a shared dictionary of weight factors for neural networks. Experiments on MNIST, FashionMNIST, and CIFAR-10 demonstrate WAFFLe's significant improvement to local test performance and fairness while simultaneously providing an extra layer of security.

The idea that the type of reset and rule-making you suggest is possible, or even could have been possible if different choices had been made in the past, is an illusion. A very simple illustration of this: look through all the articles you can find on PS that talk about "regluation" of different aspects of tech and biotech (and there are quite a few), and you will notice that literally *all of them* will pontificate that *something* should be done, or occasionally outline in broad-brush terms *what* should be done, but *not a single one* will ever suggest *how* it can be done, as in, laws that are enforcable and policable, how the penalties would work, etc. Instead, it's always all very Jean Luc Picard: "Make it happen, make it so". And it's not just writers on PS; as someone who has been following this debate for a very long time, I haven't found anyone put forward ideas about regulation of tech that can demonstrably work. If bodies do eventually get to the point of passing laws, they are of "Cookie Consent" variety - misdirected and futile nonsense.

South Korean robotics company Robotis and the Open Source Robotics Foundation (OSRF) announced the TurtleBot 3 at ROSCon last year in Seoul. We got to see a wide variety of prototypes, but Robotis was still in the middle of figuring out exactly what TurtleBot 3 was going to look like and what hardware it would include. The company told us at the time that they wanted the robot to be as open, modular, and customizable as possible, and we've been waiting excitedly to see what they came up with. Today, Robotis is finally ready to share the brand-new TurtleBot 3 with the world. And, surprise, there are actually two TurtleBot 3 models: Burger and Waffle, so named because that's kind of what each of them looks like, if you're willing to stretch your imagination a bit: A few weeks ago, Robotis shipped us test units of the two models, and after putting them together and playing a bit with them, we've got an in-depth review for you along with all the info about price and availability.