Weave, developed by the team at Weights and Biases, is a new open-source toolkit designed for performant, interactive data exploration. Our mission is to equip Machine Learning practitioners with the best tools to turn data into insights quickly and easily.

https://github.com/wandb/weave

In this tutorial, we will set up an interactive confusion matrix Weave panel using Weave expressions and Python):

https://github.com/wandb/weave/blob/master/examples/reference/confusion_matrix.ipynb

The method, called MAPseq (Multiplexed Analysis of Projections by Sequencing), makes it possible in a single experiment to trace the long-range projections of large numbers of individual neurons from a specific region or regions to wherever they lead in the brain—in experiments that are many times less expensive, labor-intensive and time-consuming than current mapping technologies allow.

https://www.cshl.edu/revolutionary-method-to-map-the-brain-at-single-neuron-resolution-is-successfully-demonstrated/ 

Tune-A-Video: One-Shot Tuning of Image Diffusion Models for Text-to-Video Generation. "We implement our findings into a simple yet effective method, dubbed as Tune-A-Video, for the task of One-Shot Video Generation"

https://tuneavideo.github.io/

Curso de IA fornecido pela Microsoft:

https://github.com/microsoft/ai-for-beginners 

Como a linguagem molda a maneira como pensamos

https://www.ted.com/talks/lera_boroditsky_how_language_shapes_the_way_we_think 

Nove pessoas com paralisia grave ou total voltaram a andar após receberem estimulação elétrica em um grupo específico de neurônios. Pesquisa do Instituto Federal de Tecnologia da Suíça, em Lausanne.

https://oglobo.globo.com/saude/noticia/2022/11/nove-pessoas-com-paralisia-voltaram-a-andar-apos-receberem-estimulacao-em-neuronios.ghtml 

Hacking pra ver:

Abra o terminal do navegador e execute:

$('.barreiraJornada').toggle();

$(':root').css('overflow', 'auto');

This course will teach you about Deep Reinforcement Learning from beginner to expert. It’s completely free and open-source!

https://huggingface.co/deep-rl-course/unit0/introduction

Reinventing Real-Time Video Communication with AI

NVIDIA Maxine is a suite of GPU-accelerated AI SDKs and cloud-native microservices for deploying AI features that enhance audio, video, and augmented reality effects in real time. 

https://developer.nvidia.com/maxine

https://www.youtube.com/shorts/adFuFz_nIMI

Language Model Programs for Embodied Control

Large language models (LLMs) trained on code-completion have been shown to be capable of synthesizing simple Python programs from docstrings.

https://code-as-policies.github.io/ 

Gloves that translate sign language to voice

Conheça a camiseta do MIT que te deixa invisível nas câmeras de segurança

https://olhardigital.com.br/2022/10/03/ciencia-e-espaco/conheca-a-camiseta-do-mit-que-te-deixa-invisivel-nas-cameras-de-seguranca/ 

Replit lança Ghostwriter, sistema similar ao GitHub Copilot, e que pode ser utilizada diretamente por meio do navegador: a ferramenta é capaz de criar funções e arquivos inteiros, refatorar e transformar código, solicitar explicações sobre códigos e autocompletar blocos ou funções inteiras.

https://replit.com/site/ghostwriter

GitHub - tim-hub/machine-learning-books: this is a collection of books and courses for machine learning.

https://github.com/tim-hub/machine-learning-books

Universidade Brasileira Livre

Um caminho para a educação autodidata em Ciência da Computação!

https://github.com/ossu/computer-science-br 

A robotic exoskeleton adapts to wearers to help them walk faster

The boot-like device uses machine learning to provide support for an individual with mobility problems.

https://www.technologyreview.com/2022/10/12/1061189/stanford-robotic-exoskeleton

VR headsets (Óculos de realidade aumentada).

Vídeo em:  https://twitter.com/MKBHD/status/1580917421757587456

High Fidelity Neural Audio Compression

EnCodec can reportedly compress audio 10 times smaller than the MP3 format at 64kbps with no loss in quality.

https://arstechnica.com/information-technology/2022/11/metas-ai-powered-audio-codec-promises-10x-compression-over-mp3/

https://arxiv.org/abs/2210.13438

New Video - Stable Diffusion: DALL-E 2 For Free, For Everyone!

Imagen Video: a text-conditional video generation system based on a cascade of video diffusion models. Given a text prompt, Imagen Video generates high definition videos using a base video generation model and a sequence of interleaved spatial and temporal video super-resolution models.

https://imagen.research.google/video/

https://tinytraining.mit.edu/

On-Device Training Under 256KB Memory

An AI used medical notes to teach itself to spot disease on chest x-rays

The new model, called CheXzero, can instead “learn” on its own from existing medical reports that specialists have written in natural language.

https://www.technologyreview.com/2022/09/15/1059541/ai-medical-notes-teach-itself-spot-disease-chest-x-rays/?utm_campaign=site_visitor.unpaid.engagement&utm_source=Twitter&utm_medium=tr_social

DALL·E 2 is a new AI system that can create realistic images and art from a description in natural language.

https://openai.com/dall-e-2/ 

AudioLM: a language modeling approach to audio generation. AudioLM achieves high-quality and long-term consistency, from audio only. Meaningful speech w/o text, consistent piano w/o MIDI.

https://arxiv.org/abs/2209.03143 

https://google-research.github.io/seanet/audiolm/examples/ 

Recent breakthroughs in text-to-image synthesis have been driven by diffusion models trained on billions of image-text pairs. Adapting this approach to 3D synthesis would require large-scale datasets of labeled 3D assets and efficient architectures for denoising 3D data, neither of which currently exist. In this work, we circumvent these limitations by using a pretrained 2D text-to-image diffusion model to perform text-to-3D synthesis.

https://dreamfusion3d.github.io/

Today, we’re announcing Make-A-Video , a new AI system that lets people turn text prompts into brief, high-quality video clips. Make-A-Video builds on Meta AI’s recent progress in generative technology research and has the potential to open new opportunities for creators and artists. The system learns what the world looks like from paired text-image data and how the world moves from video footage with no associated text. As part of our continued commitment to open science, we’re sharing details in a research paper and plan to release a demo experience.

https://ai.facebook.com/blog/generative-ai-text-to-video/

Inteligência Artificial não enxerga como os humanos - e isso não é bom

Deep learning models fail to capture the configural nature of human shape perception

Autores: Nicholas Baker, James H. Elder

Revista: iScience

DOI: 10.1016/j.isci.2022.104913

https://www.inovacaotecnologica.com.br/noticias/noticia.php?artigo=inteligencia-artificial-nao-enxerga-como-humanos-isso-nao-bom&id=010150220926#.YzHXZbdv9sU

AlphaTensor begins without any knowledge about the problem, Kohli explained, and then gradually learns what is happening and improves over time. “It first finds this classroom algorithm that we were taught, and then it finds historical algorithms such as Strassen’s and then at some point, it surpasses them and discovers completely new algorithms that are faster than previously.”

https://www.deepmind.com/blog/discovering-novel-algorithms-with-alphatensor

We present “AudioGen: Textually Guided Audio Generation”!

AudioGen is an autoregressive transformer LM that synthesizes general audio conditioned on text (Text-to-Audio).

Paper and Twitter Post

Elon Musk unveils humanoid ‘Optimus’ robot at Tesla’s AI Day

https://www.youtube.com/clip/UgkxphWN4CdSQuhYLbCbqX1d72xsPL3Yf0G4

https://www.theguardian.com/technology/2022/sep/30/tesla-optimus-humanoidrobot-elon-musk-ai-day