Mamba Explained
Is Attention all you need? Mamba, a novel AI model based on State Space Models (SSMs), emerges as a formidable alternative to the widely used Transformer models, addressing their inefficiency in processing long sequences.
Latest AI/ML News
10271 matching items
Leveraging Linux Internals to Supercharge Osquery Malware Detection
Using /proc to find fileless malware
What I learned from looking at 900 most popular open source AI tools
[ Hacker News discussion , LinkedIn discussion , Twitter thread ] Update (Feb 2026) : The full list of open source AI repos is hosted at Good AI List , updated daily. It’s balooned to 15K repos, and you can submit missing repos. You can also find some of them on my cool-llm-repos list on GitHub. Four years ago, I did an analysis of the open source ML ecosystem . Since then, the landscape has changed, so I revisited the topic. This time, I focused exclusively on the stack around foundation models. Data I searched GitHub using the keywords gpt , llm , and generative ai . If AI feels so overwhelming right now, it’s because it is. There are 118K results for gpt alone. To make my life easier, I limited my search to the repos with at least 500 stars. There were 590 results for llm , 531 for gpt , and 38 for generative ai . I also occasionally checked GitHub trending and social media for new repos. After MANY hours, I found 896 repos. Of these, 51 are tutorials (e.g. dair-ai/Prompt-Engineering-Guide ) and aggregated lists (e.g. f/awesome-chatgpt-prompts ). While these tutorials and lists are helpful, I’m more interested in software. I still include them in the final list, but the analysis is done with the 845 software repositories. It was a painful but rewarding process. It gave me a much better understanding of what people are working on, how incredibly collaborative the open source community is, and just how much China’s open source ecosystem diverges from the Western one. The Ne…
The Haitian Times thrives by understanding its audience, making smart financial decisions and embracing AI
Despite the challenges faced by the media industry, the Haitian Times –a print and digital newspaper catering to Haitian immigrants in the United States– has managed to not only survive but thrive by adapting to the changing needs of its audience. Through a combination of smart financial decisions, leveraging technology like AI, and deeply understanding […] The post The Haitian Times thrives by understanding its audience, making smart financial decisions and embracing AI appeared first on LatAm Journalism Review by the Knight Center .
The Haitian Times thrives by understanding its audience, making smart financial decisions and embracing AI
Despite the challenges faced by the media industry, the Haitian Times –a print and digital newspaper catering to Haitian immigrants in the United States– has managed to not only survive but thrive by adapting to the changing needs of its audience. Through a combination of smart financial decisions, leveraging technology like AI, and deeply understanding […] The post The Haitian Times thrives by understanding its audience, making smart financial decisions and embracing AI appeared first on LatAm Journalism Review by the Knight Center .
Car-GPT: Could LLMs finally make self-driving cars happen?
Exploring the utility of large language models in autonomous driving: Can they be trusted for self-driving cars, and what are the key challenges?
Do text embeddings perfectly encode text?
'Vec2text' can serve as a solution for accurately reverting embeddings back into text, thus highlighting the urgent need for revisiting security protocols around embedded data.
Predictive Human Preference: From Model Ranking to Model Routing
A challenge of building AI applications is choosing which model to use. What if we don’t have to? What if we can predict the best model for any prompt? Predictive human preference aims to predict which model users might prefer for a specific query. Human preference has emerged to be both the Northstar and a powerful tool for AI model development. Human preference guides post-training techniques including RLHF and DPO . Human preference is also used to rank AI models, as used by LMSYS’s Chatbot Arena . Chatbot Arena aims to determine which model is generally preferred. I wanted to see if it’s possible to predict which model is preferred for each query . One use case of predictive human preference is model routing. For example, if we know in advance that for a prompt, users will prefer Claude Instant’s response over GPT-4, and Claude Instant is cheaper/faster than GPT-4, we can route this prompt to Claude Instant. Model routing has the potential to increase response quality while reducing costs and latency. Another use case of predictive human preference is interpretability. Mapping out a model’s performance on different prompts can help us understand this model’s strengths and weaknesses. See section Experiment results for examples. Here’s what predictive human preference for different model pairs looks like for the prompt “ What’s the best way to cluster text embeddings? ”. The predictions were generated by my toy preference predictor. The bright yellow color for the (GPT-4,…
Evolution of Developer Productivity at Square - Part Four
Investing in reliability and test engineering
Why Doesn’t My Model Work?
Have you ever trained a model you thought was good, but then it failed miserably when applied to real world data? If so, you’re in good company.
Evolution of Developer Productivity at Square - Part Three
Effective developer tools
Are there cases where Variational Auto-Encoders (VAE's) are preferred to other techniques?
The best reason I have seen for using variational autoencoders is when dealing with sparse data. The Gaussian noise "splats" out the input distribution (see this StackExchange answer ). However, normalizing flows do the same thing, without the loss of information a VAE incurs. It feels as if VAEs are used only because everyone else is using them, and then a bunch of StackExchange posts reinforce the message that VAEs are the way to go , when they're theoretically suboptimal. I understand that normalizing flows are a little slower (at either training or inference), and more difficult to implement, but is there a theoretical reason that makes VAEs a legitimate choice?
Fine-tuning LLMs for longer context and better RAG systems
Update June 2024: Anyscale Endpoints (Anyscale's LLM API Offering) and Private Endpoints (self-hosted LLMs) are now available as part of the Anyscale Platform. Click [here](https://console.anyscale.com/?utm_source=anyscale&utm_medium=blog&utm_campaign=blog_callout&utm_content=june2024_product_update_subheading) to get started on the Anyscale platform.
Thinking about High-Quality Human Data
[Special thank you to Ian Kivlichan for many useful pointers (E.g. the 100+ year old Nature paper “Vox populi”) and nice feedback. 🙏 ] High-quality data is the fuel for modern data deep learning model training. Most of the task-specific labeled data comes from human annotation, such as classification task or RLHF labeling (which can be constructed as classification format) for LLM alignment training. Lots of ML techniques in the post can help with data quality, but fundamentally human data collection involves attention to details and careful execution. The community knows the value of high quality data, but somehow we have this subtle impression that “Everyone wants to do the model work, not the data work” ( Sambasivan et al. 2021 ).
Acuadata
Acuadata @administrador Mié, 31/01/2024 - 13:07 Las pérdidas de agua no técnicas en América Latina y el Caribe representan un desafío significativo en la gestión eficiente de los recursos hídricos. Estas pérdidas, que incluyen fugas y problemas en la distribución, contribuyen a un uso ineficiente del agua, lo que a su vez afecta la disponibilidad y la sostenibilidad de este recurso vital. Para abordar este problema, es esencial implementar estrategias integrales que combinen la tecnología avanzada con la capacitación adecuada y la gestión eficaz, con el objetivo de reducir las pérdidas de agua y garantizar un suministro sostenible para las comunidades en toda la región. Problema que se busca resolve En los países de la región de América Latina y el Caribe, se observan niveles elevados de pérdidas de agua, con aproximadamente el 38% del agua perdida antes de la facturación. En la Empresa Pública Metropolitana de Agua Potable y Saneamiento de Quito (EPMAPS), este indicador se sitúa en el 29%. EPMAPS tiene como objetivo reducir las pérdidas comerciales de agua en un 1%, mantener el nivel de multas y disminuir el número de inspecciones y personal en el equipo de monitoreo. Poblaciones que se ven afectadas por el problema Consumidores servicios de agua Respuesta actual a este problema, considerando a las instituciones relacionadas. Detectar las perdida no técnica de agua requiere una inmensa cantidad de datos, realizar este análisis de manera manual ha llevado a una baja efectivi…
Energizados
Energizados @administrador Mié, 31/01/2024 - 13:04 Para una sociedad que depende altamente de la disponibilidad, eficiencia y confiabilidad de la electricidad, se hace indispensable para las empresas eléctricas del sector tener una buena administración de la producción y distribución de energía. Uno de los grandes problemas que preocupa son las pérdidas eléctricas. En el transporte de energía, las pérdidas eléctricas son la diferencia entre la electricidad que ingresa a la red y la que es entregada para el consumo final, y son reflejo del nivel de eficiencia de la infraestructura en transmisión y distribución. El concepto de pérdidas eléctricas incluye también la electricidad entregada pero no facturada, que se traduce directamente en pérdidas financieras y sirve como indicador del desempeño operacional de las empresas eléctricas. Problema que se busca resolve El problema que se busca resolver a partir de las pérdidas eléctricas en la red de distribución de energía es la ineficiencia en la entrega y uso de la electricidad. Estas pérdidas pueden deberse al robo de energía, y resultan en la disipación de energía valiosa y en costos adicionales para las empresas de servicios públicos y los consumidores. La reducción de las pérdidas eléctricas no solo contribuye a un suministro más confiable y económico de energía, sino que también tiene un impacto positivo en la sostenibilidad ambiental al reducir la necesidad de generar energía adicional para compensar estas pérdidas. Poblacio…
ViaSegura
ViaSegura @administrador Mié, 31/01/2024 - 12:46 Uno de los mecanismos para salvar vidas en siniestros viales es la detección temprana de las fallas en la infraestructura vial que potencialmente pueden ocasionar las catástrofes. Problema que se busca resolve Actualmente, 1.35 millones de personas mueren y 50 millones son heridas en siniestros de tránsito por año en el mundo. Reducir la siniestralidad en las vías e incrementar la seguridad es posible. Los traumatismos causados por el tráfico vehicular son la principal causa de muerte a nivel global de niños y adultos jóvenes entre 5 y 29 años. Las muertes por accidentes de tránsito son una crisis de salud pública que puede evitarse y que afecta particularmente a los países de ingresos medio y bajo.Las lesiones graves o fatales derivadas de incidentes de tráfico son resultado del diseño y mantenimiento de la infraestructura vial. La seguridad en las carreteras es un enfoque clave del Plan Global para la Segunda Década de Acción para la Seguridad Vial 2021-2030 , que insta a los gobiernos y socios a implementar el Enfoque de Sistemas Seguros para reducir a la mitad las muertes y lesiones viales para el año 2030. Poblaciones que se ven afectadas por el problema Peatones, conductores y pasajeros de vehículos, entidades de logística, concesionarios, y autoridades viales. Respuesta actual a este problema, considerando a las instituciones relacionadas. El análisis de la seguridad vial de las carreteras se por parte de personal espec…
Pavimenta2
Pavimenta2 @administrador Mié, 31/01/2024 - 12:40 Pavimenta2 permite la detección de defectos en el pavimento de autopistas y carreteras, así como la categorización de señales de tráfico verticales y horizontales. La herramienta automatiza el análisis de las condiciones del pavimento utilizando inteligencia artificial (IA) y visión por computadora a partir de videos capturados por una cámara estándar montada en un vehículo. Esto permite a las autoridades de transporte evaluar eficazmente el inventario de carreteras, cruzar datos de accidentes y tiempos de viaje, y estimar los costos de mantenimiento con recursos limitados. La herramienta acorta un proceso que podría llevar varios años a unas pocas semanas de recopilación de videos y horas de procesamiento de imágenes. Problema que se busca resolve Los defectos en el pavimento representan más que una simple incomodidad; constituyen un riesgo significativo para los conductores, dando lugar a colisiones y fatalidades. Históricamente, la identificación de los defectos en las carreteras ha sido un proceso manual, caracterizado por su naturaleza que consume mucho tiempo y sus costos elevados. Analizar manualmente una red de carreteras de 10,000 km demanda aproximadamente 78 semanas y un presupuesto que supera los $3,000,000. Ante semejantes complejidades, los países de América Latina y el Caribe han estado realizando análisis parciales y poco frecuentes de sus redes viales. Poblaciones que se ven afectadas por el problema Autorida…
Evolution of Developer Productivity at Square - Part Two
Accelerating development velocity
Accept payments with Square using Next.js App Router
Build an end-to-end payment flow with Next.js Server Actions
Alerta de sepsis hospitalaria Laura
Alerta de sepsis hospitalaria Laura @administrador Jue, 25/01/2024 - 13:04 Se esta implementando una solución de inteligencia oficial en el Hospital Villa El Salvador para la identificación de sepsis utilizando algoritmos desarrollados por la empresa Laura. Problema que se busca resolve Detección temprana de sepsis hospitalaria Poblaciones que se ven afectadas por el problema Población internada en el hospital Villa El Salvador de Lima. Propuesta para solucionar dicho problema usando IA Detección temprana de sepsis con base en información de historia clínica electrónica. Avances/resultados Se está llevando acabo la implementación. Metas Finalización del piloto e inicio de análisis de datos. Principales retos en la implementación Participación del equipo del hospital dado el limitado numero de personal con el que cuentan. Principales retos de la IA identificados Acceso a información debidamente estructurada. Sector Salud País Perú Contacto fairlac@iadb.org Entidad Ejecutora MINSA Estado Uso y monitoreo
Protección trayectorias educativas
Protección trayectorias educativas @administrador Jue, 25/01/2024 - 13:00 El sistema educativo uruguayo presenta importantes desafíos para lograr trayectorias continuas, completas y exitosas de los jóvenes. A pesar de haber alcanzado la cobertura universal en primaria, un importante porcentaje de estudiantes uruguayos repite los primeros grados (13,4% en 1º grado y 6,9% en segundo grado). Estas altas tasas de repitencia generan rezago, y al llegar al 6º grado el 29% de los alumnos tienen sobreedad. Estos problemas en la educación primaria tienen impacto en la educación media (EM), donde Uruguay enfrenta tres retos: (i) la integración, retención y egreso de los jóvenes en el sistema; (ii) la calidad en términos de aprendizajes y desarrollo de competencias: y (iii) la equidad. Respecto al primer reto, aun cuando la EM es obligatoria la asistencia escolar disminuye a partir del ingreso a este nivel. En 2013 la tasa neta de matrícula (TNM) en la educación media básica (EMB) fue de 76,2%. A la dificultad que presenta el sistema para incorporar a los jóvenes se le suma el hecho de que el 25% de los estudiantes uruguayos de EMB tienen dos o más años de rezago escolar. Como resultado, la tasa de egreso de la EMB es solo del 57%. Todos los indicadores de la EMB ponen a Uruguay en comparación desfavorable respecto de países como Chile (88%) y Ecuador (73%). La situación se agrava en la educación media superior (EMS), donde se evidencian las dificultades del sistema para retener a los…
Predicción de Abandono y Reprobación Escolar
Predicción de Abandono y Reprobación Escolar @administrador Jue, 25/01/2024 - 12:55 El proyecto de construcción de algoritmos predictivos de deserción y fracaso escolar tiene como objetivo la modelación de algoritmos que predicen el riesgo de deserción y reprobación escolar para el nivel de secundaria. Problema que se busca resolve Reducir abandono escolar y tasas de repitencia. Poblaciones que se ven afectadas por el problema Estudiantes de 10-12 año (ensino medio) de Redes Estaduales Públicas Respuesta actual a este problema, considerando a las instituciones relacionadas. Acciones inorgánicas con escaso respaldo en información actualizada, para docentes, directores y niveles regionales de gestión. Propuesta para solucionar dicho problema usando IA Elaborar un sistema de alerta temprano para detectar abandono y reprobación – piloto con escuelas de secundaria de la Secretaria Estadual de Educação do Espírito Santo (Brasil), con apoyo del Instituto Unibanco. ¿Qué consideraciones de seguridad, leyes nacionales o estándares se tienen que tener en cuenta para utilizar cada fuente de información? Ley de protección de datos personales. Avances/resultados Convenios de uso de datos firmados: inicio de las tareas técnicas de desarrollo de algoritmos de predicción. Metas Concluir algoritmos de predicción. Elección de próxima red educativa. Principales retos en la implementación Definición de protocolos de intervención pedagógicos a partir de los riesgos identificados. Gobernanza. Prin…
Asignación centralizada de alumnos (Perú)
Asignación centralizada de alumnos (Perú) @administrador Jue, 25/01/2024 - 12:45 Alertas de riesgo 3 años. Se usa ML para predecir riesgo Problema que se busca resolve El problema descrito es que los usuarios interactúan con plataformas centralizadas de búsqueda y elección, como los sistemas de selección de escuelas, con información limitada y creencias sesgadas. Esto conduce a ineficiencias y, en ocasiones, a desigualdades en las asignaciones finales. También hay evidencia de que buscar opciones como escuelas es costoso y que los solicitantes enumeran demasiadas pocas opciones debido a creencias sesgadas sobre las probabilidades de asignación. Poblaciones que se ven afectadas por el problema Los estudiantes de Tacna que ingresan al sistema educativo entre pre-K y grado 1 (3-6 años) Respuesta actual a este problema, considerando a las instituciones relacionadas. El artículo propone colaborar con los Ministerios de Educación para enviar "boletines informativos" a los solicitantes durante la fase piloto de plataformas centralizadas de elección de escuelas. Estos boletines, dados a un grupo seleccionado al azar, proporcionan información sobre las opciones de escuelas disponibles para mejorar la toma de decisiones y las asignaciones. Propuesta para solucionar dicho problema usando IA El envío de estas tarjetas permite que los usuarios cuenten con más información en el momento de la elección de plataforma digital Avances/resultados Los estudiantes agregan más y mejores escuelas e…
Asignación centralizada de docentes (Ecuador)
Asignación centralizada de docentes (Ecuador) @administrador Jue, 25/01/2024 - 12:22 Quiero ser maestro Problema que se busca resolve El proyecto busca resolver las ineficiencias en el mercado laboral de docentes, causadas por una falta de información. Examina cómo las preferencias de los profesores por escuelas más cercanas a su hogar, ubicadas en áreas urbanas, con mejor infraestructura o con estudiantes de mayor ventaja socioeconómica, resultan en una alta demanda para algunas escuelas y vacantes en otras. Poblaciones que se ven afectadas por el problema Los maestros que participan del concurso Quiero Ser Maestro en Ecuador Respuesta actual a este problema, considerando a las instituciones relacionadas. El estudio pone a prueba una intervención con el fin de mejorar la asignación de empleos y la tasa de plazas ocupadas, proporcionando información más precisa a los docentes. Propuesta para solucionar dicho problema usando IA Específicamente, para informar mejor a los candidatos a docentes que participaron en el proceso de selección de Ecuador en 2021, estos últimos recibieron un informe personalizado a través de WhatsApp y correo electrónico que contenía un resumen de su solicitud. Para los candidatos cuyo riesgo estimado de no ser asignados era "alto" (por encima de un nivel de corte definido), el informe también incluía una advertencia de riesgo de no asignación y una lista de escuelas recomendadas donde tenían mayores posibilidades de asegurar una posición. Evaluamos el…
How Egyptian prop-tech startup Partment enables hassle-free 2nd home ownership
Egyptian prop-tech startup Partment is working to redefine second-home ownership and real estate investment via its “invest and experience” co-ownership platform. Founded in 2022, Partment offers users the chance to part-invest in meticulously curated properties, enabling personalised and diversified real estate portfolios. Through fractional ownership, co-owners access a set number of nights for personal use [...]
Announcing the $5M Core Africa Innovation Fund: Empowering Local Web3 Builders
Core DAO is excited to introduce the African Innovation Fund, a groundbreaking initiative to provide resources and networks to support local Web3 builders and projects across the African continent. This new fund is set to foster innovation, sustainability, accessibility and growth within the African blockchain ecosystem. If you’re building interesting web3 projects in Africa and [...]
UNDP launches “timbuktoo” financing initiative for African startups
The United Nations Development Programme (UNDP) has launched the “timbuktoo” initiative together with African countries, which it said is positioned to be “the world’s largest financing facility”, bringing catalytic and commercial capital together to support Africa’s startup ecosystem. President Paul Kagame of Rwanda, President Nana Akufo-Addo of Ghana, African Continental Free Trade Area secretariat secretary [...]
10 startups selected for Africa Tech Summit Nairobi Investment Showcase
Africa Tech Summit Nairobi has announced the 10 African tech ventures that will showcase their solutions to an audience of industry experts, investors and fellow innovators on February 14-15. Africa Tech Summit Nairobi is a leading African tech event providing insight and networking with the African tech ecosystem, bringing together tech leaders, MNOs, banks, international [...]
Egyptian e-health startup Yodawy banks $10m funding
Egyptian digital healthcare startup Yodawy, which has pioneered a pharmacy benefit management platform in the MENA region using technology, expert pharmacists, and state-of-the-art logistics, has raised US$10 million in extra funding, taking its total raised capital to US$34.5 million. Founded in 2018, Yodawy enables its partners – insurance companies, medical providers, pharmacies, and pharmaceutical/FMCG companies [...]
Third episode of new Disrupt Podcast series shines light on Africa’s ed-tech space
Disrupt Africa has released the third – and last – episode of its three-part podcast series zeroing in on the state of Africa’s ed-tech space, looking at trends, opportunities and challenges within the vital sector. Disrupt Podcast has released a number of focused series in the last couple of years, including ones on venture capital, [...]
Which Meme Coins to Buy in 2024? Top Meme Coins by Market Capitalization and New Trending Coins Including ApeMax, Dogecoin, Shiba Inu, Pepe Coin, Floki, Memecoin by 9gag, and Bonk
The meme coin landscape is experiencing a resurgence, with established players like Dogecoin and Shiba Inu maintaining their dominance in the market while new coins like Pepe and Dogwifhat experience recent price surges. These new coins, alongside the “Boost-to-Earn coin ApeMax, are quickly gaining traction because of their innovative features. Based on information gathered from [...]
5 Top Trending Altcoins to Buy Now | Can These Altcoins Explode After The Bitcoin Halving? (ApeMax, Arbitrum, Polkadot, Skale, Maverick Protocol)
The crypto market is buzzing with anticipation as some analysts predict a potential bull run in 2024 with key events like the upcoming Bitcoin halving event predicted to take place in April 2024. Amidst this excitement, many crypto enthusiasts, traders, and analysts are actively seeking the next big altcoin, extending their interests beyond the more [...]
Should Investors Get Excited About XRP’s $1.8 Billion Trading Volume? Why Crypto Whales are Turning to Launchpad Instead – The Best Passive Income Crypto in 2024
”Should you invest in an established crypto that shows high promise of winning its lawsuit? Or a new and relatively unknown altcoin quickly gaining traction for its presale that’s now live?” While Bitcoin has dipped by 2.80% in the past week, XRP has achieved a steady growth rate of 2.5% within the same timeframe. It’s [...]
5 Meme Coins on Investor’s Watch Lists for Upcoming Alt Coin Season 2024
Finding the next meme coin with the potential for a 100x return on investment can be a daunting task in the current crypto landscape. The market is saturated with numerous meme coins, making it challenging to identify promising projects that stand out. However, there are strategies to improve your odds of success, such approaches many [...]
Generation configurations: temperature, top-k, top-p, and test time compute
ML models are probabilistic. Imagine that you want to know what’s the best cuisine in the world. If you ask someone this question twice, a minute apart, their answers both times should be the same. If you ask a model the same question twice, its answer can change. If the model thinks that Vietnamese cuisine has a 70% chance of being the best cuisine and Italian cuisine has a 30% chance, it’ll answer “Vietnamese” 70% of the time, and “Italian” 30%. This probabilistic nature makes AI great for creative tasks. What is creativity but the ability to explore beyond the common possibilities, to think outside the box? However, this probabilistic nature also causes inconsistency and hallucinations. It’s fatal for tasks that depend on factuality. Recently, I went over 3 months’ worth of customer support requests of an AI startup I advise and found that ⅕ of the questions are because users don’t understand or don’t know how to work with this probabilistic nature. To understand why AI’s responses are probabilistic, we need to understand how models generate responses, a process known as sampling (or decoding). This post consists of 3 parts. Sampling : sampling strategies and sampling variables including temperature, top-k, and top-p. Test time compute : increasing the compute allocated to inference, e.g. sampling multiple outputs, to help improve a model’s performance. Structured outputs : how to get models to generate outputs in a certain format. Sampling Given an input, a neural networ…
Block size in subsampling and bootstrap for time series
I have a dependent variable, a time series of 80 periods (discrete decisions). I am doing maximum likelihood estimation with 10 parameters. Now I want to get the standard error or confidence interval of the estimates of these 10 parameters. One feature of my likelihood function is that the decisions is determined by all the history of $x$ , and the weight of past $x$ decreases geometrically: $x_t+\rho x_{t-1}+\rho^2 x_{t-2}$ ... where $\rho$ is one of the parameters needed to be estimated. So I am thinking that moving block bootstrapping perhaps is not suitable to sustain the data structure, and I should use subsampling. But subsampling of a given block size leaves me a very few subsamples. For example, if I choose a block size of 40, I get only 41 subsamples. Should I concern about it? Is it sufficient? Can I use multiple block sizes to construct a confidence interval? Is there any other alternatives that I could use to get standard error or confidence interval?
Deep learning for single-cell sequencing: a microscope to see the diversity of cells
On the the pivotal role that Deep Learning has played as a key enabler for advancing single-cell sequencing technologies.
How does one annotate overlapping objects in instance segmentation?
As I struggle to find any literature online about this, I wanted to ask about it here so that others could learn. My question is inspired by a Yolo GitHub issue . In this example we have 2 objects, here a plate and an egg, with one object being inside the other one. The question is how to annotate the plate (aka the outer object). Annotate full contour of outer object. Some pixels belong to 2 classes. Make a little bridge to the inner object so that the contour of the outer object excludes the inner object. This question arose while using Yolo but can be extended to other instance segmentation models. Any more information regarding good practices in instance segmentation is more than welcome.
Comment on The Future of Customer Support: How AI is Boosting Capacity, Performance, and Satisfaction by AI Worthy
Great share Kate. To be honest, I've never really thought there was much potential for AI customer service, however after using ChatGPTs new voice chat and just how natural it sounds, I was totally blown away! -Will
Comment on State-Of-The-Art Approaches to Attribution in Marketing by Bay tech media
In the realm of digital marketing, attribution methodologies have undergone significant advancements. State-of-the-art approaches include Multi-Touch Attribution (MTA) for holistic channel tracking, Algorithmic Attribution leveraging machine learning for precise credit assignment, Cross-Device Attribution capturing interactions across devices, Incrementality Testing to gauge true marketing impact, and AI-Powered Attribution for deep data analysis. Bay Tech Media implements these cutting-edge methods, empowering businesses with accurate insights to refine and optimize their marketing strategies effectively.
Comment on The Point Of Play: How Games Make AI Smarter by Mark H Black
Nowadays there are so many platforms and different technologies that make my work process easier, but even so, I still want to do nothing from time to time.
Comment on How To Get Started With Graph Machine Learning by Alex
Quite a detailed and interesting article, thanks for such a detailed description
Comment on Top AI & ML Research Trends For 2020 by Red Carson
Thanks for sharing this post
Canonical correlation analysis - loadings vs coefficients
I'm trying to wrap my head around how to interpret the results of CCA. I've got a fairly deep understanding of OLS regression, and I've read a lot of helpful CCA explainers like this one by @ttnphns . However, I'm still struggling with one particular aspect of the logic of what one apparently does with the CCA results. I'll unpack below, using terminology from the R package CCA to refer to different elements. In particular, I understand that the math of CCA treats X and Y identically, i.e., this is correlation, not regression. But, in a situation where Y is logically downstream of X, and where Y comprises multiple theoretically independent outcomes, the idea of using the ycoef s, which are essentially regression coefficients specifying the linear combination of y s that produce a given yscore , and which, like OLS regression, reflect the joint influence of the given y and all the other y s , doesn't make sense to me. Again, I understand the math of how a given yscore is derived, and how that is reflected in the ycoef s. What I don't like is the idea of reporting how the various y s 'contributed to' constructing this synthetic latent variable in a regression sense, because in reality, all the y s arose independently—or, more in keeping with the logic of CCA, they were all driven by some set of latent variables. What makes sense to me would be to report the xcoef s alongside the corr.Y.yscores , that is, the coefficients for how each x relates to a given xscore , and the loadi…
Salmon in the Loop
On fish counting – a complex sociotechnical problem in a field that is going through the process of digital transformation.
Multiple linear regression with possibly non-independent explanatory variables
For a given household for which I have many years of historical data, I want to predict the home gas consumption (heating) with a few variables among: date gas min_temp max_temp mean_temp relative_humidity absolute_humidity other_column 2023-01-01 5.8 m^3 -3.0°C 2.3°C -1.2°C 79 % 4 g/m^3 ... 2023-01-02 4.8 m^3 2.0°C 4.2°C 2.3°C 82 % 4.5 g/m^3 ... ... I could do a multiple linear regression for $$\rm{gas\ consumption} = \beta_0 + \beta_1 \rm{min\ temp} + \beta_2 \rm{max\ temp} + \beta_3 \rm{mean\ temp} + ... + \varepsilon,$$ but since many of these variables are not independent of each other (and maybe nearly collinear), doing a standard multiple linear regression might give bad results (for example with some negative $\beta_i$ where it shouldn't). Which better solution can we use? PCA + multiple linear regression (PCR) or PLS or something else? Note: I'd like to avoid using all 3 (min, max, mean) temp, if possible. How can we evaluate the loss if using using only 1 temperature variable (the best fit among the 3) instead of the 3 variables?
A new old kind of R&D lab
Answer.AI is a new kind of AI R&D lab which creates practical end-user products based on foundational research breakthroughs.
Transformers: Cross Attention Tensor Shapes During Inference Mode
Using the "classic" transformer model describing in "Attention is All You Need", I'm struggling to understand how the Encoder output is used by the Decoder during cross attention while in inference mode, specifically how the actual matrix multiplication can happen. During training mode, everything makes sense to me: The Encoder outputs a tensor of shape (B, T, C) where B = batch_size, T = max_tokens, and C = d_model = embedding dimension size. This is passed to the Decoder and changed to shape (B, T, T) through the scaled dot product mechanism (will call this tensor A ) A is multiplied by the Decoder's value tensor of shape (B, T, HS) where HS = depth = head size. This multiplication is possible because the shapes of the tensors comply (B, T, T) @ (B, T, HS) --> (B, T, HS) . But in inference mode we start with a Decoder value tensor that will only have a token length of 1 , so a tensor shape of (1, 1, HS) , where T != 1, and then expand the sequence from there. So, during the cross attention step with the Encoder, how can A with shape (1, T, T) be multiplied with (1, 1, HS) ? Clearly, I'm missing something pretty big here, so any help would be much appreciated!