Cred­it: Whit­ing School of Engi­neer­ing Johns Hop­kins researchers have devel­oped an effi­cient new method to turn blur­ry images into clear, sharp ones. Called Pro­gres­sive­ly Deblur­ring Radi­ance Field (PDRF), this approach deblurs images 15 times faster than pre­vi­ous meth­ods while also achiev­ing bet­ter results on both syn­thet­ic and real scenes. “Often­times, images are blur­ry because aut­o­fo­cus…

Cred­it: Das Shar­ma et al The inte­gra­tion of elec­tron­ic chips in com­mer­cial devices has sig­nif­i­cant­ly evolved over the past decades, with engi­neers devis­ing var­i­ous inte­gra­tion strate­gies and solu­tions. Ini­tial­ly, com­put­ers con­tained a cen­tral proces­sor or cen­tral pro­cess­ing unit (CPU), con­nect­ed to mem­o­ry units and oth­er com­po­nents via tra­di­tion­al com­mu­ni­ca­tion path­ways, known as front-side-bus (FSB) inter­faces.…

Cred­it: Tara Win­stead from Pex­els Researchers in Den­mark are har­ness­ing arti­fi­cial intel­li­gence and data from mil­lions of peo­ple to help antic­i­pate the stages of an indi­vid­u­al’s life all the way to the end, hop­ing to raise aware­ness of the tech­nol­o­gy’s pow­er, and its per­ils. Far from any mor­bid fas­ci­na­tions, the cre­ators of life2vec want to…

Raidar detects machine-gen­er­at­ed text by cal­cu­lat­ing rewrit­ing mod­i­fi­ca­tions. This illus­tra­tion shows the char­ac­ter dele­tion in red and the char­ac­ter inser­tion in orange. Human-gen­er­at­ed text tends to trig­ger more mod­i­fi­ca­tions than machine-gen­er­at­ed text when asked to be rewrit­ten. Cred­it: Yang and Von­drick labs Com­put­er sci­en­tists at Colum­bia Engi­neer­ing have devel­oped a trans­for­ma­tive method for detect­ing AI-gen­er­at­ed…

Research fields con­tribut­ing to ShELL. Cred­it: Nature Machine Intel­li­gence (2024). DOI: 10.1038/s42256-024–00800‑2 Lead­ing com­put­er sci­en­tists from around the world have shared their vision for the future of arti­fi­cial intelligence—and it resem­bles the capa­bil­i­ties of Star Trek char­ac­ter “The Borg.” Experts from the likes of Lough­bor­ough Uni­ver­si­ty, MIT, and Yale say we are set to see…

The Umwelt inter­face. A) The data, visu­al, and audio tabs of the edi­tor. B) The editor’s fields tab, where users spec­i­fy field def­i­n­i­tions and encod­ings. C) The view­er, where users ana­lyze data with inter­ac­tive mul­ti­modal data rep­re­sen­ta­tions. Cred­it: arX­iv (2024). DOI: 10.48550/arxiv.2403.00106 A grow­ing num­ber of tools enable users to make online data rep­re­sen­ta­tions, like…

The image on the left shows a nor­mal chest X‑ray, where­as the one on the right shows lungs with pneu­mo­nia opac­i­ty (Bre­viglieri, 2019). Cred­it: Fron­tiers in Com­put­er Sci­ence (2024). DOI: 10.3389/fcomp.2023.1286657 Pneu­mo­nia, an infec­tion in the lungs that caus­es dif­fi­cul­ty breath­ing, is most com­mon­ly diag­nosed through chest X‑rays. Typ­i­cal­ly, those chest X‑rays are read by…

Schemat­ic view of (top) cur­rent­ly used HAMR and (bot­tom) three-dimen­­sion­al mag­net­ic record­ing sys­tems. In the three-dimen­­sion­al mag­net­ic record­ing sys­tem, the Curie tem­per­a­ture of each record­ing lay­er dif­fers by about 100 K and data are writ­ten to each lay­er by adjust­ing the laser pow­er. Cred­it: Acta Mate­ri­alia (2024). DOI: 10.1016/j.actamat.2024.119869 Research groups from NIMS, Sea­gate Tech­nol­o­gy,…

Large-scale lan­guage mod­els enable AI agents to evolve var­i­ous types of per­son­al­i­ties in social inter­ac­tions. Cred­it: Reiko Mat­sushi­ta Researchers in Japan have effec­tive­ly devel­oped a diverse range of per­son­al­i­ty traits in dia­logue AI using a large-scale lan­guage mod­el (LLM). Using the pris­on­er’s dilem­ma from game the­o­ry, Pro­fes­sor Takaya Ari­ta and Asso­ciate Pro­fes­sor Rei­ji Suzu­ki from…

Dr. Pei Wang teach­es a arti­fi­cial gen­er­al intel­li­gence class at Tem­ple Uni­ver­si­ty in Philadel­phia, Thurs­day, Feb. 1, 2024. Main­stream AI research “turned away from the orig­i­nal vision of arti­fi­cial intel­li­gence, which at the begin­ning was pret­ty ambi­tious,” said Wang. Cred­it: AP Photo/Matt Rourke There’s a race under­way to build arti­fi­cial gen­er­al intel­li­gence, a futur­is­tic vision…

Qui­et-STaR. We visu­al­ize the algo­rithm as applied dur­ing train­ing to a sin­gle thought. We gen­er­ate thoughts, in par­al­lel, fol­low­ing all tokens in the text (think). The mod­el pro­duces a mix­ture of its next-token pre­dic­tions with and with­out a thought (talk). We apply REINFORCE, as in STaR, to increase the like­li­hood of thoughts that help the…

With their DMD method, MIT researchers cre­at­ed a one-step AI image gen­er­a­tor that achieves image qual­i­ty com­pa­ra­ble to Sta­bleD­if­fu­sion v1.5 while being 30 times faster. Cred­it: Illus­tra­tion by Alex Shipps/MIT CSAIL using six AI-gen­er­at­ed images devel­oped by researchers. In our cur­rent age of arti­fi­cial intel­li­gence, com­put­ers can gen­er­ate their own “art” by way of dif­fu­sion…

Cred­it: AI-gen­er­at­ed image Large lan­guage mod­els, such as those that pow­er pop­u­lar arti­fi­cial intel­li­gence chat­bots like Chat­G­PT, are incred­i­bly com­plex. Even though these mod­els are being used as tools in many areas, such as cus­tomer sup­port, code gen­er­a­tion, and lan­guage trans­la­tion, sci­en­tists still don’t ful­ly grasp how they work. In an effort to bet­ter under­stand…

The pro­posed algo­rithm com­bines vari­a­tion­al sched­ul­ing with post-pro­cess­ing to achieve near-opti­­mal solu­tions to com­bi­na­to­r­i­al opti­miza­tion prob­lems with con­straints with­in the oper­a­tion time of quan­tum com­put­ers. Cred­it: Tat­suhiko Shi­rai / Wase­da Uni­ver­si­ty Com­bi­na­to­r­i­al opti­miza­tion prob­lems (COPs) have appli­ca­tions in many dif­fer­ent fields such as logis­tics, sup­ply chain man­age­ment, machine learn­ing, mate­r­i­al design and drug dis­cov­ery, among…

Each dot in this fig­ure is a con­vo­lu­tion­al neur­al net­work mod­el with the ener­gy con­sump­tion on hor­i­zon­tal axis and per­for­mance on ver­ti­cal axis. Con­ven­tion­al­ly, mod­els are select­ed only based on their performance—without tak­ing their ener­gy con­sump­tion into account—resulting in mod­els in the red ellipse. This work enables prac­ti­tion­ers to select mod­els from the green ellipse…